Search

Learning-based translation between MRI contrasts involves supervised deep models trained using high-quality source- and target-contrast images derived from fully-sampled acquisitions, which might be difficult to collect under limitations on scan costs or time. To facilitate curation of training sets, here we introduce the first semi-supervised model for MRI contrast translation (ssGAN) that can be trained directly using undersampled k-space data. To enable semi-supervised learning on undersampled data, ssGAN introduces novel multi-coil losses in image, k-space, and adversarial domains. The multi-coil losses are selectively enforced on acquired k-space samples unlike traditional losses in single-coil synthesis models. Comprehensive experiments on retrospectively undersampled multi-contrast brain MRI datasets are provided. Our results demonstrate that ssGAN yields on par performance to a supervised model, while outperforming single-coil models trained on coil-combined magnitude images. It also outperforms cascaded reconstruction-synthesis models where a supervised synthesis model is trained following self-supervised reconstruction of undersampled data. Thus, ssGAN holds great promise to improve the feasibility of learning-based multi-contrast MRI synthesis.

This paper introduces semi-universal perturbations that bridge the gap between specific and universal adversarial perturbations. The original idea is to craft a specific perturbation by choosing it among a set of $L$ universal perturbations. We propose to jointly learn the perturbations of this set to maximize the chances to attack each example by allowing it to choose its own perturbation. To do so, we derive an algorithm, with convergence guarantees under Lipschitz continuity assumptions. Semi-universal perturbations offer a better flexibility, interpretability and diversity, confirmed by our experiments. Additionally, we provide a generalization bound on the abilities of the perturbations to attack new examples.

Supervised reconstruction models are characteristically trained on matched pairs of undersampled and fully-sampled data to capture an MRI prior, along with supervision regarding the imaging operator to enforce data consistency. To reduce supervision requirements, the recent deep image prior framework instead conjoins untrained MRI priors with the imaging operator during inference. Yet, canonical convolutional architectures are suboptimal in capturing long-range relationships, and priors based on randomly initialized networks may yield suboptimal performance. To address these limitations, here we introduce a novel unsupervised MRI reconstruction method based on zero-Shot Learned Adversarial TransformERs (SLATER). SLATER embodies a deep adversarial network with cross-attention transformers to map noise and latent variables onto coil-combined MR images. During pre-training, this unconditional network learns a high-quality MRI prior in an unsupervised generative modeling task. During inference, a zero-shot reconstruction is then performed by incorporating the imaging operator and optimizing the prior to maximize consistency to undersampled data. Comprehensive experiments on brain MRI datasets clearly demonstrate the superior performance of SLATER against state-of-the-art unsupervised methods.

Le reti neurali profonde (DNNs) sono potenti modelli predittivi, che superano le capacità umane in una varietà di task. Imparano sistemi decisionali complessi e flessibili dai dati a disposizione e raggiungono prestazioni eccezionali in molteplici campi di apprendimento automatico, dalle applicazioni dell'intelligenza artificiale, come il riconoscimento di immagini, parole e testi, alle scienze più tradizionali, tra cui medicina, fisica e biologia. Nonostante i risultati eccezionali, le prestazioni elevate e l’alta precisione predittiva non sono sufficienti per le applicazioni nel mondo reale, specialmente in ambienti critici per la sicurezza, dove l'utilizzo dei DNNs è fortemente limitato dalla loro natura black-box. Vi è una crescente necessità di comprendere come vengono eseguite le predizioni, fornire stime di incertezza, garantire robustezza agli attacchi avversari e prevenire comportamenti indesiderati. Anche le migliori architetture sono vulnerabili a piccole perturbazioni nei dati di input, note come attacchi avversari: manipolazioni malevole degli input che sono percettivamente indistinguibili dai campioni originali ma sono in grado di ingannare il modello in predizioni errate. In questo lavoro, dimostriamo che tale fragilità è correlata alla geometria del manifold dei dati ed è quindi probabile che sia una caratteristica intrinseca delle predizioni dei DNNs. Questa condizione suggerisce una possibile direzione al fine di ottenere robustezza agli attacchi: studiamo la geometria degli attacchi avversari nel limite di un numero infinito di dati e di pesi per le reti neurali Bayesiane, dimostrando che, in questo limite, sono immuni agli attacchi avversari gradient-based. Inoltre, proponiamo alcune tecniche di training per migliorare la robustezza delle architetture deterministiche. In particolare, osserviamo sperimentalmente che ensembles di reti neurali addestrati su proiezioni casuali degli input originali in spazi basso-dimensionali sono più resistenti agli attacchi. Successivamente, ci concentriamo sul problema dell'interpretabilità delle predizioni delle reti nel contesto delle saliency-based explanations. Analizziamo la stabilità delle explanations soggette ad attacchi avversari e dimostriamo che, nel limite di un numero infinito di dati e di pesi, le interpretazioni Bayesiane sono più stabili di quelle fornite dalle reti deterministiche. Confermiamo questo comportamento in modo sperimentale nel regime di un numero finito di dati. Infine, introduciamo il concetto di attacco avversario alle sequenze di amminoacidi per protein Language Models (LM). I modelli di Deep Learning per la predizione della struttura delle proteine, come AlphaFold2, sfruttano le architetture Transformer e il loro meccanismo di attention per catturare le proprietà strutturali e funzionali delle sequenze di amminoacidi. Nonostante l'elevata precisione delle predizioni, perturbazioni biologicamente piccole delle sequenze di input, o anche mutazioni di un singolo amminoacido, possono portare a strutture 3D sostanzialmente diverse. Al contempo, i protein LMs sono insensibili alle mutazioni che inducono misfolding o disfunzione (ad esempio le missense mutations). In particolare, le predizioni delle coordinate 3D non rivelano l'effetto di unfolding indotto da queste mutazioni. Pertanto, esiste un'evidente incoerenza tra l'importanza biologica delle mutazioni e il conseguente cambiamento nella predizione strutturale. Ispirati da questo problema, introduciamo il concetto di perturbazione avversaria delle sequenze proteiche negli embedding continui dei protein LMs. Il nostro metodo utilizza i valori di attention per rilevare le posizioni degli amminoacidi più vulnerabili nelle sequenze di input. Le mutazioni avversarie sono biologicamente diverse dalle sequenze di riferimento e sono in grado di alterare in modo significativo le strutture 3D. Deep Neural Networks (DNNs) are powerful predictive models, exceeding human capabilities in a variety of tasks. They learn complex and flexible decision systems from the available data and achieve exceptional performances in multiple machine learning fields, spanning from applications in artificial intelligence, such as image, speech and text recognition, to the more traditional sciences, including medicine, physics and biology. Despite the outstanding achievements, high performance and high predictive accuracy are not sufficient for real-world applications, especially in safety-critical settings, where the usage of DNNs is severely limited by their black-box nature. There is an increasing need to understand how predictions are performed, to provide uncertainty estimates, to guarantee robustness to malicious attacks and to prevent unwanted behaviours. State-of-the-art DNNs are vulnerable to small perturbations in the input data, known as adversarial attacks: maliciously crafted manipulations of the inputs that are perceptually indistinguishable from the original samples but are capable of fooling the model into incorrect predictions. In this work, we prove that such brittleness is related to the geometry of the data manifold and is therefore likely to be an intrinsic feature of DNNs’ predictions. This negative condition suggests a possible direction to overcome such limitation: we study the geometry of adversarial attacks in the large-data, overparameterized limit for Bayesian Neural Networks and prove that, in this limit, they are immune to gradient-based adversarial attacks. Furthermore, we propose some training techniques to improve the adversarial robustness of deterministic architectures. In particular, we experimentally observe that ensembles of NNs trained on random projections of the original inputs into lower dimensional spaces are more resilient to the attacks. Next, we focus on the problem of interpretability of NNs’ predictions in the setting of saliency-based explanations. We analyze the stability of the explanations under adversarial attacks on the inputs and we prove that, in the large-data and overparameterized limit, Bayesian interpretations are more stable than those provided by deterministic networks. We validate this behaviour in multiple experimental settings in the finite data regime. Finally, we introduce the concept of adversarial perturbations of amino acid sequences for protein Language Models (LMs). Deep Learning models for protein structure prediction, such as AlphaFold2, leverage Transformer architectures and their attention mechanism to capture structural and functional properties of amino acid sequences. Despite the high accuracy of predictions, biologically small perturbations of the input sequences, or even single point mutations, can lead to substantially different 3d structures. On the other hand, protein language models are insensitive to mutations that induce misfolding or dysfunction (e.g. missense mutations). Precisely, predictions of the 3d coordinates do not reveal the structure-disruptive effect of these mutations. Therefore, there is an evident inconsistency between the biological importance of mutations and the resulting change in structural prediction. Inspired by this problem, we introduce the concept of adversarial perturbation of protein sequences in continuous embedding spaces of protein language models. Our method relies on attention scores to detect the most vulnerable amino acid positions in the input sequences. Adversarial mutations are biologically diverse from their references and are able to significantly alter the resulting 3D structures.

Generative adversarial networks (GANs) have become increasingly powerful, generating mind-blowing photorealistic images that mimic the content of datasets they have been trained to replicate. One recurrent theme in medical imaging, is whether GANs can also be as effective at generating workable medical data, as they are for generating realistic RGB images. In this paper, we perform a multi-GAN and multi-application study, to gauge the benefits of GANs in medical imaging. We tested various GAN architectures, from basic DCGAN to more sophisticated style-based GANs, on three medical imaging modalities and organs, namely: cardiac cine-MRI, liver CT, and RGB retina images. GANs were trained on well-known and widely utilized datasets, from which their FID scores were computed, to measure the visual acuity of their generated images. We further tested their usefulness by measuring the segmentation accuracy of a U-Net trained on these generated images and the original data. The results reveal that GANs are far from being equal, as some are ill-suited for medical imaging applications, while others performed much better. The top-performing GANs are capable of generating realistic-looking medical images by FID standards, that can fool trained experts in a visual Turing test and comply to some metrics. However, segmentation results suggest that no GAN is capable of reproducing the full richness of medical datasets.

Accepted submission for NORDES Doctoral Consortium 2023 (to be published in the Extended Abstracts) My research concerns Designing for Contestable Systems1. Looking into how we can negotiate and contextualise automated decision systems at the front-end and address power imbalances between people and the systems which increasingly intervene to take decisions about our futures. Coming from a background of anarchistic self-organisation, my first inspiration came from the Right to Repair movement. Understanding our bad interactions with automated systems as a form of brokenness provides a less abstract perspective on our relationships with these distant processes and to think of them as repairable. Taking the Repair Café as an example of grassroots activism, I proposed the concept of the Contestation Café (Collins, 2022). A community-based space where the public could come to share experiences about unfair and inappropriate treatment by algorithmic systems and to find support and ways to contest and redesign them. Gatherings around these concerns can create the kind of critical mass which results in systemic change, like the recent backtracking with Apple and John Deere (Purdy, 2022), where large companies were forced to recant their tactics and allow their users to repair and have ownership over how their products are used. In the same way, we can challenge how our data is used and how automated decisions can be repaired, negotiated with and understood. While it is helpful for the general public to have the tools to "punch upwards" at oppressive systems, it is also necessary to provide designers with ways to pass down affordances for these users to contest with their designs. This is where my research intersects with the Nordes theme. Neither the designer or the user should have the sole responsibility for making technology fairer - rather, I believe it should be an ongoing and participatory activity through the interface and through shared conversation. This is where I perceive the blank space to be - in a lack of connection between the primary stakeholders, the users, the designers and the system itself. New design approaches are needed to bridge this space and fill in the blanks. I am exploring agonism as a way to value the struggle between the user, designer and system - to encourage discourse, reconfiguration and to view design as the journey towards solutions that can always be challenged and will never become concrete (Miessen et al., 2012). Within this, I am inhabiting the role of The Fixer, a character who might "know and see different things - indeed different worlds - than the better known figures of "designer" and "user"" (Jackson, 2014), and who could act as a 'contester-on-behalf- of' the bleeding edge of society and for those from more marginalised communities who are often the first affected by new technologies. My next steps will be to look for examples of agonistic design through history, consult with a broad sample of stakeholders involved in contemporary design and technology, and to define a set of values for this design approach which brings design and repair together and opens up a space for contestability, redesign and discourse as an ongoing condition of design., {"references":["Purdy (2022) Apple and John Deere Shareholder Resolutions Demand They Explain Their Bad Repair Policies | iFixit News [Internet]. Available from: [Accessed 14 February 2023].","Collins, R. (2022) The Contestation Café - Prototyping An Agonistic Space. Available from: [Accessed 6 December 2022].","Jackson (2014) Rethinking Repair (MITPress2014).","Miessen, M., Mouffe, C. & Mroué, R. (2012) The space of agonism. Berlin, Sternberg Press."]}

The Onion Router (Tor) is used by adversaries and warfighters alike to encrypt session information and gain anonymity on the internet. Since its creation in 2002, Tor has gained popularity by terrorist organizations, human traffickers, and illegal drug distributors who wish to use Tor services to mask their identity while engaging in illegal activities. Fingerprinting attacks assist in thwarting these attempts. Website fingerprinting (WF) attacks have been proven successful at linking a user to the website they have viewed over an encrypted Tor connection. With consumer video streaming traffic making up a large majority of internet traffic and sites like YouTube remaining in the top visited sites in the world, it is just as likely that adversaries are using videos to spread misinformation, illegal content, and terrorist propaganda. Video fingerprinting (VF) attacks look to use encrypted network traffic to predict the content of encrypted video sessions in closed- and open-world scenarios. This research builds upon an existing dataset of encrypted video session data and use statistical analysis to train a machine-learning classifier, using deep fingerprinting (DF), to predict videos viewed over Tor. DF is a machine learning technique that relies on the use of convolutional neural networks (CNN) and can be used to conduct VF attacks against Tor. By analyzing the results of these experiments, we can more accurately identify malicious video streaming activity over Tor. Civilian Approved for public release. Distribution is unlimited.

The level of realization of the right to a fair trial is one of the crucial indicators of democracy in any state. In order to ensure this right, all the minimum standards deriving from it must be clearly understood by law enforcement agencies and their practice must meet these standards. As equality of arms, the right to a fair trial, is not directly enshrined in the text of Article 6 of the European Convention on Human Rights (hereafter - the Convention) and is of implicit character; the issues like its essence, content and the way it should manifest in practice are open for discussion. For this reason, the focus on those issues is highly relevant. The aim of this article is, with reference to the case law of the European Court of Human Rights (hereafter - ECHR) and the modern doctrine based on this right, to explain the role of this principle and the essence of its mutual relations with the other elements of the right to a fair trial. Selected case law of ECHR bears great interest compared with other decisions and is discussed in the form of empirical materials of the study. From the doctrinal materials, interpretation of Article 6 of the Convention and theoretical sources related to the European standards in the criminal procedure are also analyzed. The article exercises methods of dialectical comprehension; they are determinism, induction, deduction, case studies and methods of law interpretation. As a result of the study, a unique doctrinal commentary has been obtained in the context of adversariality and impartial and independent court principles of the concept of equality of arms, as well as, interaction of the minimum rights of persons subject to criminal prosecution, guaranteed by the Convention.

The Contestation Manifesto and its associated paraphernalia are artefacts from a speculative, near-future community action known as the Contestation Café. Being one in a series of research through design projects on contestation, the Contestation Café is a critical, yet also practical, design intervention rooted in how the act of repair has moved through the physical into the digital, and the shared values therein. Tracing the history of electronic repair - from the early valve radios, the invention of the transistor, microchips, programmable devices and through to IoT, connected devices and the “fluid assemblages” that emerge - we can see the shared values of the physical act of repair with the more intangible act of contestation in the digital world of algorithmic systems. Overlaying the values and tactics of the Right to Repair movement with the emerging concerns around data-driven systems we find ourselves examining the Repair Café as a potential model for community contestation and the construction of publics. The Repair Café started in Amsterdam in 2009 and has since spread to 35 countries with over 1700 instances of these cafés. The Repair Café is not intended to be a place where you bring your broken appliances for someone else to repair - rather, it is a place where you learn how to repair and recycle your own products and devices, and - more importantly - a place where you simply learn that things can be fixed rather than thrown out. In a similar spirit, the Contestation Café would be a place for those who feel mistreated by automated decision systems, AIs and algorithms, to bring their broken interactions and their unfair decisions to learn how to contest, push back and reclaim their agency and autonomy. Rather than Repairers, the Contestation Café has a panel of Fixers - people who inhabit the space between designers and users, with a particular knowledge of these systems and how to map and navigate them - who are there to share their experience and knowledge, and to guide the user in the ways of contestation and to become Fixers of their own futures. Although the Contestation Café is a speculative concept based on research into the shared values of contestation and repair, it has developed a life of its own through the process of imagining how it would work and what it would look like. Through the act of writing a manifesto for contestation, the space has manifested itself in the present and is waiting only for the fixers and public to arrive. Plans are already in progress for its first real instance and all of the imagined artefacts presented here will inform this process. https://contestationcafe.org/

Jurors’ evaluations of expert testimony should focus on the quality of experts’ testimony to inform decisions. However, when experts are highly paid and testify frequently, jurors are more likely to see them as ‘hired guns’, biasing their decisions and ratings of the expert’s credibility. In this study, I examine whether using court appointed experts can attenuate the effects of these heuristics on jurors’ decisions by varying who the expert testifies for, the quality of the testimony, and the presence of the heuristics. Thus, I will test whether using court-appointed experts can eliminate the biasing hired gun effect on jurors’ decisions.

As the internet continues to be populated with new devices and emerging technologies, the attack surface grows exponentially. Technology is shifting towards a profit-driven Internet of Things market where security is an afterthought. Traditional defending approaches are no longer sufficient to detect both known and unknown attacks to high accuracy. Machine learning intrusion detection systems have proven their success in identifying unknown attacks with high precision. Nevertheless, machine learning models are also vulnerable to attacks. Adversarial examples can be used to evaluate the robustness of a designed model before it is deployed. Further, using adversarial examples is critical to creating a robust model designed for an adversarial environment. Our work evaluates both traditional machine learning and deep learning models' robustness using the Bot-IoT dataset. Our methodology included two main approaches. First, label poisoning, used to cause incorrect classification by the model. Second, the fast gradient sign method, used to evade detection measures. The experiments demonstrated that an attacker could manipulate or circumvent detection with significant probability., Comment: MDPI Mach. Learn. Knowl. Extr. 2021, 3(2), 333-356; https://www.mdpi.com/2624-800X/1/2/14

International audience; Magnetic Resonance Imaging (MRI) acquisition is performed sequentially in the spatial-frequency domain (kspace) and involves several views of the object/subject. To accelerate the acquisition, k-space lines are often undersampled on a cartesian grid. Parallel imaging reconstruction algorithms are then applied to recover unseen lines. We consider this super-resolution problem in k-space to further reduce the acquisition time with deep learning and to reduce the costs associated with this expensive medical imaging technology. Because the sensors are specific to anatomical regions and experimental setups, it pushes toward learning a reconstruction model on a per-image basis, i.e. for every scan and image. Here, we propose an extension of state-of-the-art MRI reconstruction methods where the super-resolution task in k-space is solved with a convolutional neural network, and where an adversarial strategy using a patch discriminator in image space is used to reach higher undersampling rates. Both parts are trained in a one-shot learning setting. It is demonstrated both using simulated and in vivo brain experiments that this combined approach provides enhanced image quality for undersampling rates larger than the ones used in a clinical routine on multi-slice 2D T2-weighted imaging sequences, making the approach readily applicable as an alternative reconstruction strategy in MRI systems for 2D parallel imaging, and which could be further extended to accelerate 3D imaging sequences.

Cataloged from PDF version of article. Thesis (Master's): Bilkent University, Department of Electrical and Electronics Engineering, İhsan Doğramacı Bilkent University, 2022. Includes bibliographical references (leaves 45-71). Supervised reconstruction models are characteristically trained on matched pairs of undersampled and fully-sampled data to capture an MRI prior, along with supervision regarding the imaging operator to enforce data consistency. To reduce supervision requirements, the recent deep image prior framework instead conjoins untrained MRI priors with the imaging operator during inference. Yet, canonical convolutional architectures are suboptimal in capturing long-range relationships, and priors based on randomly initialized networks may yield suboptimal performance. To address these limitations, this thesis introduces a novel unsupervised MRI reconstruction method based on zero-Shot Learned Adversarial TransformERs (SLATER). SLATER embodies a deep adversarial network with cross-attention transformers to map noise and latent variables onto coil-combined MR images. During pre-training, this unconditional network learns a high-quality MRI prior in an unsupervised generative modeling task. During inference, a zero-shot reconstruction is then performed by incorporating the imaging operator and optimizing the prior to maximize consistency to undersampled data. Comprehensive experiments on brain MRI datasets clearly demonstrate the superior performance of SLATER against state-of-the-art unsupervised methods. by Yılmaz Korkmaz M.S.

Fair classification has become an important topic in machine learning research. While most bias mitigation strategies focus on neural networks, we noticed a lack of work on fair classifiers based on decision trees even though they have proven very efficient. In an up-to-date comparison of state-of-the-art classification algorithms in tabular data, tree boosting outperforms deep learning (Zhang et al. in Expert Syst Appl 82:128–150, 2017). For this reason, we have developed a novel approach of adversarial gradient tree boosting. The objective of the algorithm is to predict the output Y with gradient tree boosting while minimizing the ability of an adversarial neural network to predict the sensitive attribute S. The approach incorporates at each iteration the gradient of the neural network directly in the gradient tree boosting. We empirically assess our approach on four popular data sets and compare against state-of-the-art algorithms. The results show that our algorithm achieves a higher accuracy while obtaining the same level of fairness, as measured using a set of different common fairness definitions.

Generative adversarial models with convolutional neural network (CNN) backbones have recently been established as state-of-the-art in numerous medical image synthesis tasks. However, CNNs are designed to perform local processing with compact filters, and this inductive bias compromises learning of contextual features. Here, we propose a novel generative adversarial approach for medical image synthesis, ResViT, that leverages the contextual sensitivity of vision transformers along with the precision of convolution operators and realism of adversarial learning.} ResViT's generator employs a central bottleneck comprising novel aggregated residual transformer (ART) blocks that synergistically combine residual convolutional and transformer modules. Residual connections in ART blocks promote diversity in captured representations, while a channel compression module distills task-relevant information. A weight sharing strategy is introduced among ART blocks to mitigate computational burden. A unified implementation is introduced to avoid the need to rebuild separate synthesis models for varying source-target modality configurations. Comprehensive demonstrations are performed for synthesizing missing sequences in multi-contrast MRI, and CT images from MRI. Our results indicate superiority of ResViT against competing CNN- and transformer-based methods in terms of qualitative observations and quantitative metrics.

The database of faces containing sensitive information is at risk of being targeted by unauthorized automatic recognition systems, which is a significant concern for privacy. Although there are existing methods that aim to conceal identifiable information by adding adversarial perturbations to faces, they suffer from noticeable distortions that significantly compromise visual perception, and therefore, offer limited protection to privacy. Furthermore, the increasing prevalence of appearance anxiety on social media has led to users preferring to beautify their faces before uploading images. In this paper, we design a novel face database protection scheme via beautification with chaotic systems. Specifically, we construct the adversarial face with better visual perception via beautification for each face in the database. In the training, the face matcher and the beautification discriminator are federated against the generator, prompting it to generate beauty-like perturbations on the face to confuse the face matcher. Namely, the pixel changes produced by face beautification mask the adversarial perturbations. Moreover, we use chaotic systems to disrupt the order of adversarial faces in the database, further mitigating the risk of privacy leakage. Our scheme has been extensively evaluated through experiments, which show that it effectively defends against unauthorized attacks while also yielding good visual results.

En el presente trabajo se realiza un estudio histórico, analítico, sistemático y jurisprudencial del Ministerio Público en Colombia y su participación en el proceso penal, como interviniente especial, lo cual permite establecer la necesidad y/o utilidad de su intervención al interior del proceso penal con tendencia acusatoria, bajo el entendido de que estamos en presencia de un proceso penal, adaptado a nuestras especiales realidades, distando mucho de parecerse a un sistema acusatorio puro, lo cual permitirá al lector comprender el grado de incidencia que tiene su participación en el proceso penal y la utilidad de la figura en estudio frente al tema de la defensa de los intereses de la sociedad, de los derechos humanos y del debido proceso, conceptos que han sido caros en el sistema procesal penal colombiano y que se han querido proteger a ultranza de la arbitrariedad en el proceso penal, razones que fueron entendidas por el legislador, cuando previó su participación en el proceso penal, como importante para el mantenimiento de las garantías y derechos de los ciudadanos, quienes se ven expuestos a la pugna entre eficientismo y garantismo. In the present work, a historical, analytical, systematic and jurisprudential study of the Public Ministry in Colombia and its participation in the criminal process, as a special intervener, is carried out, which allows establishing the need and usefulness of its intervention within the criminal process. with an accusatory tendency, under the understanding that we are in the presence of a criminal process, adapted to our special realities, far from resembling a pure accusatory system, which will allow the reader to understand the degree of incidence that their participation has in the process criminal law and the usefulness of the figure under study in relation to the issue of defending the interests of society, human rights and due process, concepts that have been expensive in the Colombian criminal procedure system and that have been sought to be protected at all arbitrariness in the criminal process, reasons that were understood by the legislator, when he foresaw his participation in the criminal process, as important for the maintenance of the guarantees and rights of citizens, who are exposed to conflict between efficiency and guarantee. Maestría Magíster en Justicia y Tutela de los Derechos con énfasis en Ciencias Penales y Criminológicas (Modalidad de Investigación)

We present SNEAD, an extension of an algorithmic framework by McKelvey (1998) under which one can reduce the problem of finding Nash equilibria in finite games to a tractable unified optimization problem. Specifically, we adapt the method to tractably optimize continuous-strategy games and improve asymptotic running time over finite-strategy games. We further prove that convergence of our algorithm to zero loss guarantees a Nash equilibrium in quasiconcave games. Experimentally, our algorithm achieves state of the art performance on a variety of games, even in the face of high-dimensional, complex payoff functions. Finally, we provide supplementary theoretical frameworks which further adapt SNEAD to locate symmetric equilibria or subgame-perfect equilibria with novel algorithmic efficiency.

The vulnerability of the high-performance machine learning models implies a security risk in applications with real-world consequences. Research on adversarial attacks is beneficial in guiding the development of machine learning models on the one hand and finding targeted defenses on the other. However, most of the adversarial attacks today leverage the gradient or logit information from the models to generate adversarial perturbation. Works in the more realistic domain: decision-based attacks, which generate adversarial perturbation solely based on observing the output label of the targeted model, are still relatively rare and mostly use gradient-estimation strategies. In this work, we propose a pixel-wise decision-based attack algorithm that finds a distribution of adversarial perturbation through a reinforcement learning algorithm. We call this method Decision-based Black-box Attack with Reinforcement learning (DBAR). Experiments show that the proposed approach outperforms state-of-the-art decision-based attacks with a higher attack success rate and greater transferability.

Τα τελευταία χρόνια, πληθώρα μοντέλων Βαθιάς Μάθησης έχουν προταθεί στη βιβλιογραφία για την επίλυση των προβλημάτων της ταξινόμησης εικόνων και ανίχνευσης αντικειμένων σε εικόνες, επιτυγχάνοντας μάλιστα αποτελέσματα υψηλής ακρίβειας σε διαφορετικά σύνολα δεδομένων. Ωστόσο, η απόδοσή τους μπορεί να επηρεαστεί σημαντικά από τις αντιπαραθετικές επιθέσεις. Ο σκοπός των αντιπαραθετικών επιθέσεων είναι η επίδραση τους στις αρχικές εικόνες εισόδου με τρόπο ώστε ενώ οι εικόνες που προκύπτουν να είναι σχεδόν πανομοιότυπες με τις αρχικές ως προς το ανθρώπινο μάτι, να παραπλανούν τα αποτελέσματα των αντίστοιχων ταξινομητών. Αυτές οι μέθοδοι μπορούν να διακριθούν σε επιθέσεις βασισμένες στην εικόνα εισόδου και στις καθολικές επιθέσεις. Η πλειοψηφία αυτών των μεθόδων χρησιμοποιούν προσθετικό θόρυβο στην εικόνα εισόδου. Στη τρέχουσα διπλωματική διατριβή, αντιμετωπίζουμε το πρόβλημα της δημιουργίας καθολικών αντιπαραθετικών επιθέσεων, ως μία εκτίμηση μετασχηματισμών, εξετάζοντας την απλούστερη περίπτωση όπου ο μετασχηματισμός είναι γραμμικός. Ως εκ τούτου, μπορούμε να ενοποιήσουμε δύο μορφές εισαγόμενης διαταραχής (πολλαπλασιαστική και προσθετική) και υποστηρίζουμε ότι το πολλαπλασιαστικό μέρος του μετασχηματισμού μπορεί να είναι επίσης αποτελεσματικό, οδηγώντας σε λιγότερο θορυβώδη αντιπαραθετικά παραδείγματα. Οι δημιουργούμενες αντιπαραθετικές επιθέσεις είναι αναστρέψιμες αλλά δεν είναι εύκολα ανακτήσιμες. Πειραματικά αποτελέσματα παρουσιάζονται για την αξιολόγηση της προτεινόμενης επίθεσης σε σύγκριση με δύο υπάρχουσες μεθόδους της βιβλιογραφίας. Η αξιολόγηση και η σύγκριση των μεθόδων λαμβάνει υπόψη τόσο την ακρίβεια ταξινόμησης πριν και μετά την εφαρμογή των μαθησιακών επιθέσεων, καθώς και τις μετρήσεις ποιότητας του Μέσου Τετράγωνου Σφάλματος και του Δείκτη Δομικής Ομοιότητας. Τα πειράματα διεξάγονται και παρουσιάζονται σε πέντε σύνολα δεδομένων (MNIST, Fashion-MNIST, CIFAR-10, CIFAR-100, STL-10) που χρησιμοποιούνται συνήθως, για την αξιολόγηση μεθόδων, στον τομέα της όρασης υπολογιστών. Τα αποτελέσματα καταδεικνύουν ότι η προτεινόμενη μέθοδος μπορεί να οδηγήσει σε παρόμοια ή και καλύτερα αποτελέσματα από τις μεθόδους της βιβλιογραφίας, ενώ λόγω της πολλαπλασιαστικής φύσης της είναι δύσκολο να ανακτηθεί. Επιπλέον, παρουσιάζονται πειράματα σχετικά με τον αριθμό των ζευγών από αρχικές και επιτιθέμενες εικόνες που απαιτούνται για την ανάκτηση της αντιπαραθετικής επίθεσης. Η προτεινόμενη μέθοδος μπορεί να βρει εφαρμογή τόσο για την αξιολόγηση της ευρωστίας των μοντέλων βαθιάς μάθησης, όσο και για τη διατήρηση της ιδιωτικότητας ατόμων από συστήματα αναγνώρισης προσώπων., In recent years many Deep Learning models have been proposed in the literature for image classification and object detection tasks, which achieve high accuracy results in different datasets. However, their performance can be significantly affected by adversarial attacks. The scope of adversarial attacks is to perturb images in a way that they are almost identical to the human eye while being misled by the respective classifiers. Those methods can be distinguished in image-specific and universal attacks. Most of those attacks use additive noise on the input image. In this thesis, we address the problem of universal adversarial attack generation, in deep neural network classification, as a transformation estimation one. We examine the simplest case where the transformation is linear. As such, we can unify two forms of introduced perturbation (multiplicative and additive) and we are arguing that the multiplicative part of the transformation can be also effective, leading in less noisy adversarial examples. The generated adversarial perturbations are revertible but not easily attainable. Experimental results are presented for the proposed attack along with a comparison with two existing methods of literature. The evaluation and the comparison of the methods consider both the classification accuracy before and after the application of the learned attacks, along with the quality metrics of Mean Square Error (MSE) and the Structural Similarity Index Measure (SSIM). Experiments are conducted and presented in five datasets (i.e. MNIST, Fashion-MNIST, CIFAR-10, CIFAR-100, STL-10) that are commonly used, for methods evaluation, in the field of computer vision. The results demonstrate that the proposed method can lead to similar or even better results than the methods of literature, while due to their multiplicative nature it is difficult to be attained. Moreover, experiments are presented regarding the number of pairs of input and perturbed images that are needed for obtaining the adversarial attack. The proposed method can find application both for evaluating the robustness of deep learning models, as well as for preserving human privacy in face recognition systems.

It is possible to extract authorization from the procedural system about the application of enforcement measures against nonparties as a result of the duties imputed to all without distinction, of being subject to the effects of the decision and of the general effective power structured in CPC/15. This reality had already been verified by an important written work by Sérgio Cruz Arenhart at the beginning of the last century, trend that was reinvigorated with the current procedural law. In contrast, no one can be harmed without due processo of law, so participation in an adversarial process is elementary for what is being analyzed. É possível extrair do sistema processual a autorização para a aplicação de medidas de efetivação em face de terceiros, isso em decorrência dos deveres imputados a todos indistintamente, da sujeição aos efeitos da decisão e do poder geral de efetivação estruturado no CPC/15. Essa realidade já havia sido constatada por importante trabalho escrito de Sérgio Cruz Arenhart no início do século passado, tendência que se revigorou com a atual lei processual. Em contrapartida, ninguém pode ser prejudicado sem o devido processo legal, de modo que a participação em contraditório é elementar para o que se analisa.

Adversarial examples are inputs to a machine learning system that result in an incorrect output from that system. Attacks launched through this type of input can cause severe consequences: for example, in the field of image recognition, a stop signal can be misclassified as a speed limit indication.However, adversarial examples also represent the fuel for a flurry of research directions in different domains and applications. Here, we give an overview of how they can be profitably exploited as powerful tools to build stronger learning models, capable of better-withstanding attacks, for two crucial tasks: fake news and social bot detection., To appear on IEEE Internet Computing. `Accepted manuscript' version

International audience; L’espace du débat parlementaire à la Chambre des Communes Cette communication a pour objectif d’explorer les potentialités du cadre normatif dans l’espace des débats parlementaires à la Chambre des Communes à Westminster.Les débats parlementaires à la Chambre des Communes sont régis par des règles et des protocoles stricts. En effet, les membres de la Chambre doivent observer des règles de bonne conduite afin d’éviter tout débordement. Ce cadre institutionnel (et constitutionnel), à première vue très rigide, concerne à la fois les contraintes physiques au sein de la Chambre mais également les contraintes d’ordre discursif. Sandra Harris, en référence à la Chambre de Communes, parle de ‘communauté de pratique’ indiquant qu’il s’agit bien là d’un espace social à part entière. Les députés passent par un processus de socialisation (bien souvent par le biais de différents rituels) et se fondent dans le cadre normatif de la Chambre presque instinctivement. Parfois, il est plus difficile pour certains membres de s’approprier cet espace, par exemple les femmes députées qui sont quelques fois forcées d’entrer (par une sorte d’acte d’acculturation) dans le cadre confrontationnel et souvent agressif des débats. Les normes et les coutumes sont devenues des marqueurs idéologiques, la chambre se divise en différents territoires qu’il est nécessaire de défendre. Le parti au pouvoir est toujours à la droite du Speaker alors que l’Opposition se trouve à sa gauche. Passer physiquement “de l’autre côté” est un acte de rébellion très fort.Dans cette lutte, les règles deviennent un instrument de pouvoir. Celui ou celle qui saura le mieux se servir du cadre (physique et/ou normatif) sortira vainqueur du débat. Dans cette perspective, le cadre devient un instrument de pouvoir. On ne cherche pas à le détruire, mais plutôt à en tirer le maximum de profit (politique) ou encore, à l’instar de Ian Blackford (qui se fait exclure de la Chambre, symbole fort), à en sortir pour intégrer un autre espace, un autre système, celui de Holyrood. D’autres tentent d’élargir leur auditoire en usant de stratégies de contournement des règles, par exemple en trouvant des parades pour s’adresser plus directement (ou plus intimement) à leurs mandants, instaurant une dynamique dedans/dehors.

A lo largo del presente trabajo el autor se concentra, en primer lugar, en algunas ideas esenciales de Nils Christie para lograr una reducción de dolor y una resolución distinta a los conflictos que llegan al sistema penal. Luego compara el desarrollo y la estrategia de este autor con la de Eugenio Raúl Zaffaroni en nuestra región Latinoamericana. Y, finalmente, delinea algunas reflexiones sobre por qué considera que hoy en día, y más precisamente en provincias que ya cuentan con procesos adversariales-acusatorios, resulta sumamente necesario retomar a Christie y resucitar sus ideas para construir -o rediseñar- un sistema que dé respuestas distintas que las que damos siempre con nuestras prácticas cotidianas. Fil: Barrios, Nicanor. Universidad Nacional de la Patagonia "san Juan Bosco". Facultad de Ciencias Jurídicas; Argentina. Consejo Nacional de Investigaciones Científicas y Técnicas. Centro Científico Tecnológico Conicet - Centro Nacional Patagónico. Instituto Patagónico de Ciencias Sociales y Humanas; Argentina

À partir d'études de cas centrées sur des demandes de séparation de couples par le moyen judiciaire de l'adultère masculin ou du concubinage aux xvie et premier xviie siècles, cet article mène une comparaison entre deux types de juridictions compétentes pour ce chef, l'une laïque et royale, l'autre ecclésiastique ; et deux types de procédures, l'une accusatoire et l'autre inquisitoire. Pour les premières, l'accusation d'adultère renvoie à plusieurs séries de manquements du mari qui sont d'ordre sexuels, génératifs, économiques et patrimoniaux mais aussi, moraux et politiques. Dans les secondes, la mise en défaut est d'ordre formaliste et de contrevenue à l'ordre sacramentel, à l'exclusion d'autres considérations. Les cas soumis à l'examen judiciaire montrent de leurs côtés des pratiques de couple non conformes aux normes officielles. Based on case studies centered on couples' requests for separation by the judicial means of male adultery or cohabitation in the sixteenth and early seventeenth centuries, this article compares two types of competent jurisdictions for this charge, one secular and royal, the other ecclesiastical ; and two types of procedures, one accusatory and the other inquisitorial. For the former, the accusation of adultery refers to several series of failures by the husband of a sexual, generative, economic or patrimonial order but also, moral or political. In the second case, the accusation is of a formal nature and a breach of the sacramental order, to the exclusion of other considerations. The cases submitted for judicial review show , the practices of couples that did not conform to official norms.

The article gives a brief historical description of the role of the court in the collection of evidence. The arguments in favor of strengthening the role of the court for the recovery of evidence are presented. In particular, the Author substantiates the conclusion that it is necessary to impose on the court the obligation to independently demand evidence in civil proceedings. The Author comes to the conclusion that strengthening the role of the court in the process of proof in civil proceedings are dictated by the requirements due to the many flaws in civil procedural law, the unequal social status of participants in the process, making decisions not in accordance with actual circumstances of the case.

We propose an online algorithm for sequential learning in the contextual multiarmed bandit setting. Our approach is to partition the context space and, then, optimally combine all of the possible mappings between the partition regions and the set of bandit arms in a data-driven manner. We show that in our approach, the best mapping is able to approximate the best arm selection policy to any desired degree under mild Lipschitz conditions. Therefore, we design our algorithm based on the optimal adaptive combination and asymptotically achieve the performance of the best mapping as well as the best arm selection policy. This optimality is also guaranteed to hold even in adversarial environments since we do not rely on any statistical assumptions regarding the contexts or the loss of the bandit arms. Moreover, we design an efficient implementation for our algorithm using various hierarchical partitioning structures, such as lexicographical or arbitrary position splitting and binary trees (BTs) (and several other partitioning examples). For instance, in the case of BT partitioning, the computational complexity is only log-linear in the number of regions in the finest partition. In conclusion, we provide significant performance improvements by introducing upper bounds (with respect to the best arm selection policy) that are mathematically proven to vanish in the average loss per round sense at a faster rate compared to the state of the art. Our experimental work extensively covers various scenarios ranging from bandit settings to multiclass classification with real and synthetic data. In these experiments, we show that our algorithm is highly superior to the state-of-the-art techniques while maintaining the introduced mathematical guarantees and a computationally decent scalability. IEEE

El principio de igualdad de armas es un mandato esencial que consiste en que “cada parte debe tener una oportunidad razonable para presentar su caso en condiciones que no la pongan en desventaja con respecto a su oponente”1. Esta figura, a pesar de su connotada y evidente importancia, no ha sido objeto de estudio profundo y serio en la doctrina y la jurisprudencia colombianas. Lo anterior ha permitido un uso irresponsable y equivocado de este concepto, y es de ahí que nace la necesidad de este artículo, el cual tiene por objeto, sin pretender ser exhaustivo, dar una mayor claridad sobre este mandato de op­timización, empezando por revisar sus antecedentes; pasando por el análisis de cada uno de sus elementos esenciales; la referencia a algunas categorías desarrolladas en la jurisdicción internacional que son de suma importancia para alcanzar un entendimiento íntegro de dicho principio; el señalamiento de las diferentes posiciones sobre su fundamento, definición y alcance; y una diferenciación con otros derechos y principios con los que suele confundirse. De esta manera, se sentarán las bases para que, posteriormente, sea posible abrir un debate organizado acerca de esta figura que haga viable un ejercicio optimizado y responsable de ella. The principle of equality of arms is an essential optimization require­ment, which requires that “each party must be afforded a reasonable opportunity to present his case under conditions that do not place him at a disadvantage vis-à-vis his opponent”. This procedural figure, notwithstanding, has not been studied deeply and seriously in Colombian doctrine and jurisprudence despite its evident importance. The aforementioned situation has allowed an irrespon­sible and mistaken use of this concept and that is the reason why this article arises, which aims, without pretending to be exhaustive, to give a greater clarity about this optimization requirement, starting with a study about its antecedents, going through the analysis of its essential features, the reference to different positions about its basis, definition, content and scope, its differentiation with other rights and principles and the reference to some categories developed at the international jurisdiction that are relevant to achieve a full understanding of it. Subsequently, it will be possible to open an organized debate on this procedural figure in order to make viable an optimized and responsible use of this concept.

Risk analysis is an essential methodology for cybersecurity as it allows organizations to deal with cyber threats potentially affecting them, prioritize the defense of their assets, and decide what security controls should be implemented. Many risk analysis methods are present in cybersecurity models, compliance frameworks, and international standards. However, most of them employ risk matrices, which suffer shortcomings that may lead to suboptimal resource allocations. We propose a comprehensive framework for cybersecurity risk analysis, covering the presence of both intentional and nonintentional threats and the use of insurance as part of the security portfolio. A simplified case study illustrates the proposed framework, serving as template for more complex problems., Horizon 2020 Framework Programme. Grant Number: 740920 Ministerio de Economía y Competitividad. Grant Number: MTM2014‐56949‐C3‐1‐R AXA Research Fund. Grant Number: AXA‐ICMAT Chair on Adversarial Risk Analysis European Cooperation in Science and Technology. Grant Number: IS1304

�� ������������ ���������������������� ���������������������� ������������ �������������� ���������������������� ������������������ �������������� �� ������������������ ���������������������������������� ����������������, ���������������������������� �������������� ���������������������������������� ��������������������������������. ������������������������, ������ ���������������������������� �������������� ���������������������������������� �������������������������������� ������������������������ ���� ������������������ (�������������������� ������������������������������ ������������������������ ��������, ������������������������ ��������������������, �������������� ������������������ ������������ �� ������������), �� �������������� ������������������ ��������������������������, ���������������������������� ��������������������-���������������� ���������� �������������������������� ������������������������ ������������������, �� �������������� �������������������� ��������������������. �������������� �������������������� �������������� �������� ������������������ �������������������������� ���������������� ���������������������� ��������������, �������������������� ���� ���������������������������� �������� ������������������������������ ������������������������ ����������������. ���������������� �������������������� ������������������������ �������������������� ���������������������������������� ������������ ���������������������������������� ��������������������������������, �������� �������������� �������������� �� ������, ������ ���������������������������� ������ ������������������������������ ����������������, ������������������������������ ������ ���������������� ������������ �� ���������������� ���������� ����������, ��.��. ��������, �������������������������� ���� ���������������� ������������������������ ��������, ���������������������� ���������� ������ ������������������������ ������������ �������������������������������� ������ ���� �������������������� ������������������ ���������������� �������������������� �� ������������������, �� ������������ �� ��������, ���������������������� �������������������������� �������������� �� ���������������������� �������� ������������������������ �������������������������������� ����������������������������. ���� ��������, ������ �������������������� �������������� ���� ������������������ �������������� ������������������������ ���������������������������� ��������������, ���������������������������� ������������ �������� �� ������������ �������� ������ �������������������� ������������������������������ ������ ��������������������, the article presents a critical analysis of the position of supporters of the judicial approach to understanding the administrative process about the procedural nature of administrative proceedings. It is proved that the procedural nature of administrative legal proceedings is determined not by the subject (the branch affiliation of the material norms that fix the permits that have become the subject of a ���dispute about law���), but by the method of legal regulation represented by the civil law type of regulation of public relations, in which permits dominate. The channel of implementation of this type of legal regulation is a civil process based on an adversarial type of jurisdictional legal process. The shortcomings of the model of administrative legal proceedings fixed by the Russian legislation are revealed, the essence of which is that the adversarial type of jurisdictional process intended for judicial protection in the litigation cases of the plaintiff, i.e. the person who suffered from the actions of an official, is also used when considering other administrative cases on the implementation of certain power requirements to the defendant, namely, the person who violated the established prohibitions and restrictions, or committed an administrative offense. In fact, for the implementation of sanctions for violation of the ban, a procedural procedure is used that ensures the protection of the rights and freedoms of a person when implementing the permissions granted to him

In this paper, we compare different audio signal representations, including the raw audio waveform and a variety of time-frequency representations, for the task of audio synthesis with Generative Adversarial Networks (GANs). We conduct the experiments on a subset of the NSynth dataset. The architecture follows the benchmark Progressive Growing Wasserstein GAN. We perform experiments both in a fully non-conditional manner as well as conditioning the network on the pitch information. We quantitatively evaluate the generated material utilizing standard metrics for assessing generative models, and compare training and sampling times. We show that complex-valued as well as the magnitude and Instantaneous Frequency of the Short-Time Fourier Transform achieve the best results, and yield fast generation and inversion times. The code for feature extraction, training and evaluating the model is available online., 5 pages, 1 figure, 5 tables, to be published in European Signal Processing Conference (EUSIPCO)

Deep learning has proven to be a major leap in machine learning, allowing completely new problems to be solved. While flexible and powerful, neural networks have the disadvantage of being large and demanding high performance from the devices on which they are run. In order to deploy neural networks on more, and simpler, devices, techniques such as quantization, sparsification and tensor decomposition have been developed. These techniques have shown promising results, but their effects on model robustness against attacks remain largely unexplored. In this thesis, Universal Adversarial Perturbations (UAP) and the Fast Gradient Sign Method (FGSM) are tested against VGG-19 as well as versions of it compressed using 8-bit quantization, TensorFlow’s float16 quantization, and 8-bit and 4-bit single layer quantization as introduced in this thesis. The results show that UAP transfers well to all quantized models, while the transferability of FGSM is high to the float16 quantized model, lower to the 8-bit models, and high to the 4-bit SLQ model. We suggest that this disparity arises from the universal adversarial perturbations’ having been trained on multiple examples rather than just one, which has previously been shown to increase transferability. The results also show that quantizing a single layer, the first layer in this case, can have a disproportionate impact on transferability. Examensarbetet är utfört vid Institutionen för teknik och naturvetenskap (ITN) vid Tekniska fakulteten, Linköpings universitet

This short note presents an analysis of the principle of adversarial embedding (a.k.a. adv-emb) in steganography as presented in [4]. By analyzing the covariance matrix of the stego signal of quantized JPEG coefficients (i.e. the signal added to the JPEG Cover image to create the Stego image), we highlight the fact that the stego signal exhibits correlations between coefficients. These weak correlations are within the same block (intra-block correlations) of between adjacent blocks (inter-block correlations). The correlation patterns are similar to the patterns of correlations analyzed on the sensor noise in the DCT domain (see [3]). However, if in [3] these correlations have been shown to favor continuities between blocks, the correlations induced by adversarial embedding are on opposite sign and they code discontinuities between blocks (see Figure 1b). An experiment consisting in generating first a cover image using J-Uniward [1] and then computing an adversarial signal using PGD [2] exhibits very similar patterns than with adversarial embedding (see figure 1c). We consequently presume that, as pictured in Figure 2, the adversarial signal is generated in order to compensate the blocking artifacts created by the embedding. This would explain the sign of the correlations. Our rationale is that the proportion β of the DCT coefficients used produce the adverse stego signal in adv-emb conveys the signal whose role is to try to remove the blocking artifacts generated by additive schemes such as J-Uniward.

This paper addresses the problem of semi-supervised transfer learning with limited cross-modality data in remote sensing. A large amount of multi-modal earth observation images, such as multispectral imagery (MSI) or synthetic aperture radar (SAR) data, are openly available on a global scale, enabling parsing global urban scenes through remote sensing imagery. However, their ability in identifying materials (pixel-wise classification) remains limited, due to the noisy collection environment and poor discriminative information as well as limited number of well-annotated training images. To this end, we propose a novel cross-modal deep-learning framework, called X-ModalNet, with three well-designed modules: self-adversarial module, interactive learning module, and label propagation module, by learning to transfer more discriminative information from a small-scale hyperspectral image (HSI) into the classification task using a large-scale MSI or SAR data. Significantly, X-ModalNet generalizes well, owing to propagating labels on an updatable graph constructed by high-level features on the top of the network, yielding semi-supervised cross-modality learning. We evaluate X-ModalNet on two multi-modal remote sensing datasets (HSI-MSI and HSI-SAR) and achieve a significant improvement in comparison with several state-of-the-art methods.

Diplomski rad obuhvaća kratku teoretsku razradu vrsta učenja modela u području dubokog učenja, te opsežnu teoretsku razradu metode polunadziranog učenja pod nazivom Polunadzirano učenje s virtualnim neprijateljskim primjerima. Predstavlja regularizacijsku tehniku učenja s virtualnim neprijateljskim primjerima koja potiče glatkoću decizijske granice zahtijevajući da neoznačeni primjeri imaju što sličniju predikciju kao i njihove perturbirane inačice. Predstavlja pojam virtualne neprijateljske perturbacije, te funkciju lokalne glatkoće razdiobe koje predstavljaju ključne pojmove u metodi. Provedeni su eksperimenti nad skupovima podataka MNIST, SVHN te CIFAR-10. Razvijen je postupak polunadziranog učenja klasifikacijskog modela zasnovan na virtualnim neprijateljskim primjerima. The masters thesis includes a short theoretical elaboration of the types of model training in the field of deep learning and an extensive theoretical elaboration of the method of semi-supervised learning called Semi-supervised learning with virtual adversarial examples. It represents a regularization technique with virtual adversarial examples that encourages the smoothness of the decision boundary by requiring unlabeled examples to have as similar prediction as possible to their perturbed versions. It represents the term of virtual adversarial perturbation and the local smoothness distribution function that represent key terms in the method. Experiments were performed on the data sets MNIST, SVHN and CIFAR-10. A process of semi-supervised training of the classification model based on virtual adversarial examples has been developed.

Motivacija za problem izgradnje suparničkih primjera, opisana je u uvodnom poglavlju. Obrada prirodnog jezika i koncepti suparničkih napada ukratko su opisani. Glavni cilj rada je razvoj raznolikih napada u području prirodnog jezika. Naglasak je stavljen na učinkovite napade i na metode koje mogu očuvati strukturalnu i semantičku cjelovitost teksta. Razvijeni su novi pristupi napada. Oslanjaju se na procjenu značajnosti pojedinih riječi u tekstu, kao i na intenzivno korištenje modela za leksičku zamjenu te ručno rađenih reprezentacija riječi. Provedeni su brojni eksperimenti kako bi se procijenile i usporedile izvedbe pojedinih suparničkih napada. Rad je zaključen kratkih pregledom mogućih obrambenih mehanizama i uvidima koji su stečeni na temelju dobivenih rezultata. Motivation for this assignment, crafting adversarial examples for textual machine learning models, is presented in the introductory chapter. Natural language processing and adversarial attack concepts, important for understanding the experiments, are briefly described. Main topic of this assignment is development of various attacks in the natural language domain. Emphasis is put on effective attacks and methods that can preserve structural and semantic integrity of text. Novel approaches have been designed. They rely on evaluating importance of certain words in text, along with the heavy use of lexical substitution models and custom-crafted word embeddings. A series of experiments had been conducted in order to evaluate and compare different adversarial attacks. The thesis is concluded with short overview of possible defence mechanisms and insights gathered from numerous results.

International audience; This paper addresses the problem of semi-supervised transfer learning with limited cross-modality data in remote sensing. A large amount of multi-modal earth observation images, such as multispectral imagery (MSI) or synthetic aperture radar (SAR) data, are openly available on a global scale, enabling parsing global urban scenes through remote sensing imagery. However, their ability in identifying materials (pixel-wise classification) remains limited, due to the noisy collection environment and poor discriminative information as well as limited number of well-annotated training images. To this end, we propose a novel cross-modal deep-learning framework, called X-ModalNet, with three well-designed modules: self-adversarial module, interactive learning module, and label propagation module, by learning to transfer more discriminative information from a small-scale hyperspectral image (HSI) into the classification task using a large-scale MSI or SAR data. Significantly, X-ModalNet generalizes well, owing to propagating labels on an updatable graph constructed by high-level features on the top of the network, yielding semi-supervised cross-modality learning. We evaluate X-ModalNet on two multi-modal remote sensing datasets (HSI-MSI and HSI-SAR) and achieve a significant improvement in comparison with several state-of-the-art methods.

Modern-day machine learning algorithms are vulnerable to adversarial perturbations. Before machine learning is employed in operational environments it is important for the Department of Defense to understand the vulnerabilities in adversarial domains. With an emphasis on applications to image classification, this thesis intends to investigate methods for constructing adversarial manipulations intended to fool a statistical classifier in order to establish best practices prior to employment in a mission-critical environment. The state-of-the-art method for generating adversarial perturbations utilizes an easy to obtain linear approximation of the defender’s loss function and applies change to each pixel’s value in the direction of the gradient by a pre-determined constant in order to maximize the defender’s loss function. This results in misclassification of the image. This thesis aims to improve existing methods for developing adversarial inputs to image classifiers by attempting to find new ways to make perturbations less perceptible to human vision. We develop two new perturbation methods. Color Aware attempts to constrain perturbation changes to match human perception by mapping to CIELAB color space, a color space better suited to represent human vision, while Color & Edge Aware constrains perturbation changes in visually smooth regions in images. Outstanding Thesis Captain, United States Marine Corps Approved for public release. distribution is unlimited

Training neural networks to be certifiably robust is critical to ensure their safety against adversarial attacks. However, it is currently very difficult to train a neural network that is both accurate and certifiably robust. In this work we take a step towards addressing this challenge. We prove that for every continuous function $f$, there exists a network $n$ such that: (i) $n$ approximates $f$ arbitrarily close, and (ii) simple interval bound propagation of a region $B$ through $n$ yields a result that is arbitrarily close to the optimal output of $f$ on $B$. Our result can be seen as a Universal Approximation Theorem for interval-certified ReLU networks. To the best of our knowledge, this is the first work to prove the existence of accurate, interval-certified networks., ICLR 2020

Nous étudions dans cette thèse comment les réseaux de neurones profonds peuvent être utilisés dans différents domaines de la vision artificielle. La vision artificielle est un domaine interdisciplinaire qui traite de la compréhension d’images et de vidéos numériques. Les problèmes de ce domaine ont traditionnellement été adressés avec des méthodes ad-hoc nécessitant beaucoup de réglages manuels. En effet, ces systèmes de vision artificiels comprenaient jusqu’à récemment une série de modules optimisés indépendamment. Cette approche est très raisonnable dans la mesure où, avec peu de données, elle bénéficient autant que possible des connaissances du chercheur. Mais cette avantage peut se révéler être une limitation si certaines données d’entré n’ont pas été considérées dans la conception de l’algorithme. Avec des volumes et une diversité de données toujours plus grands, ainsi que des capacités de calcul plus rapides et économiques, les réseaux de neurones profonds optimisés d’un bout à l’autre sont devenus une alternative attrayante. Nous démontrons leur avantage avec une série d’articles de recherche, chacun d’entre eux trouvant une solution à base de réseaux de neurones profonds à un problème d’analyse ou de synthèse visuelle particulier. Dans le premier article, nous considérons un problème de vision classique: la détection de bords et de contours. Nous partons de l’approche classique et la rendons plus ‘neurale’ en combinant deux étapes, la détection et la description de motifs visuels, en un seul réseau convolutionnel. Cette méthode, qui peut ainsi s’adapter à de nouveaux ensembles de données, s’avère être au moins aussi précis que les méthodes conventionnelles quand il s’agit de domaines qui leur sont favorables, tout en étant beaucoup plus robuste dans des domaines plus générales. Dans le deuxième article, nous construisons une nouvelle architecture pour la manipulation d’images qui utilise l’idée que la majorité des pixels produits peuvent d’être copiés de l’image d’entrée. Cette technique bénéficie de plusieurs avantages majeurs par rapport à l’approche conventionnelle en apprentissage profond. En effet, elle conserve les détails de l’image d’origine, n’introduit pas d’aberrations grâce à la capacité limitée du réseau sous-jacent et simplifie l’apprentissage. Nous démontrons l’efficacité de cette architecture dans le cadre d’une tâche de correction du regard, où notre système produit d’excellents résultats. Dans le troisième article, nous nous éclipsons de la vision artificielle pour étudier le problème plus générale de l’adaptation à de nouveaux domaines. Nous développons un nouvel algorithme d’apprentissage, qui assure l’adaptation avec un objectif auxiliaire à la tâche principale. Nous cherchons ainsi à extraire des motifs qui permettent d’accomplir la tâche mais qui ne permettent pas à un réseau dédié de reconnaître le domaine. Ce réseau est optimisé de manière simultané avec les motifs en question, et a pour tâche de reconnaître le domaine de provenance des motifs. Cette technique est simple à implémenter, et conduit pourtant à l’état de l’art sur toutes les tâches de référence. Enfin, le quatrième article présente un nouveau type de modèle génératif d’images. À l’opposé des approches conventionnels à base de réseaux de neurones convolutionnels, notre système baptisé SPIRAL décrit les images en termes de programmes bas-niveau qui sont exécutés par un logiciel de graphisme ordinaire. Entre autres, ceci permet à l’algorithme de ne pas s’attarder sur les détails de l’image, et de se concentrer plutôt sur sa structure globale. L’espace latent de notre modèle est, par construction, interprétable et permet de manipuler des images de façon prévisible. Nous montrons la capacité et l’agilité de cette approche sur plusieurs bases de données de référence., In the present thesis, we study how deep neural networks can be applied to various tasks in computer vision. Computer vision is an interdisciplinary field that deals with understanding of digital images and video. Traditionally, the problems arising in this domain were tackled using heavily hand-engineered adhoc methods. A typical computer vision system up until recently consisted of a sequence of independent modules which barely talked to each other. Such an approach is quite reasonable in the case of limited data as it takes major advantage of the researcher's domain expertise. This strength turns into a weakness if some of the input scenarios are overlooked in the algorithm design process. With the rapidly increasing volumes and varieties of data and the advent of cheaper and faster computational resources end-to-end deep neural networks have become an appealing alternative to the traditional computer vision pipelines. We demonstrate this in a series of research articles, each of which considers a particular task of either image analysis or synthesis and presenting a solution based on a ``deep'' backbone. In the first article, we deal with a classic low-level vision problem of edge detection. Inspired by a top-performing non-neural approach, we take a step towards building an end-to-end system by combining feature extraction and description in a single convolutional network. The resulting fully data-driven method matches or surpasses the detection quality of the existing conventional approaches in the settings for which they were designed while being significantly more usable in the out-of-domain situations. In our second article, we introduce a custom architecture for image manipulation based on the idea that most of the pixels in the output image can be directly copied from the input. This technique bears several significant advantages over the naive black-box neural approach. It retains the level of detail of the original images, does not introduce artifacts due to insufficient capacity of the underlying neural network and simplifies training process, to name a few. We demonstrate the efficiency of the proposed architecture on the challenging gaze correction task where our system achieves excellent results. In the third article, we slightly diverge from pure computer vision and study a more general problem of domain adaption. There, we introduce a novel training-time algorithm (\ie, adaptation is attained by using an auxilliary objective in addition to the main one). We seek to extract features that maximally confuse a dedicated network called domain classifier while being useful for the task at hand. The domain classifier is learned simultaneosly with the features and attempts to tell whether those features are coming from the source or the target domain. The proposed technique is easy to implement, yet results in superior performance in all the standard benchmarks. Finally, the fourth article presents a new kind of generative model for image data. Unlike conventional neural network based approaches our system dubbed SPIRAL describes images in terms of concise low-level programs executed by off-the-shelf rendering software used by humans to create visual content. Among other things, this allows SPIRAL not to waste its capacity on minutae of datasets and focus more on the global structure. The latent space of our model is easily interpretable by design and provides means for predictable image manipulation. We test our approach on several popular datasets and demonstrate its power and flexibility.

L'évaluation de la démarche humaine est l'une des composantes essentielles dans les soins de santé. Les systèmes à base de marqueurs avec plusieurs caméras sont largement utilisés pour faire cette analyse. Cependant, ces systèmes nécessitent généralement des équipements spécifiques à prix élevé et/ou des moyens de calcul intensif. Afin de réduire le coût de ces dispositifs, nous nous concentrons sur un système d'analyse de la marche qui utilise une seule caméra de profondeur. Le principe de notre travail est similaire aux systèmes multi-caméras, mais l'ensemble de caméras est remplacé par un seul capteur de profondeur et des miroirs. Chaque miroir dans notre configuration joue le rôle d'une caméra qui capture la scène sous un point de vue différent. Puisque nous n'utilisons qu'une seule caméra, il est ainsi possible d'éviter l'étape de synchronisation et également de réduire le coût de l'appareillage. Notre thèse peut être divisée en deux sections: reconstruction 3D et analyse de la marche. Le résultat de la première section est utilisé comme entrée de la seconde. Notre système pour la reconstruction 3D est constitué d'une caméra de profondeur et deux miroirs. Deux types de capteurs de profondeur, qui se distinguent sur la base du mécanisme d'estimation de profondeur, ont été utilisés dans nos travaux. Avec la technique de lumière structurée (SL) intégrée dans le capteur Kinect 1, nous effectuons la reconstruction 3D à partir des principes de l'optique géométrique. Pour augmenter le niveau des détails du modèle reconstruit en 3D, la Kinect 2 qui estime la profondeur par temps de vol (ToF), est ensuite utilisée pour l'acquisition d'images. Cependant, en raison de réflections multiples sur les miroirs, il se produit une distorsion de la profondeur dans notre système. Nous proposons donc une approche simple pour réduire cette distorsion avant d'appliquer les techniques d'optique géométrique pour reconstruire un nuage de points de l'objet 3D. Pour l'analyse de la démarche, nous proposons diverses alternatives centrées sur la normalité de la marche et la mesure de sa symétrie. Cela devrait être utile lors de traitements cliniques pour évaluer, par exemple, la récupération du patient après une intervention chirurgicale. Ces méthodes se composent d'approches avec ou sans modèle qui ont des inconvénients et avantages différents. Dans cette thèse, nous présentons 3 méthodes qui traitent directement les nuages de points reconstruits dans la section précédente. La première utilise la corrélation croisée des demi-corps gauche et droit pour évaluer la symétrie de la démarche, tandis que les deux autres methodes utilisent des autoencodeurs issus de l'apprentissage profond pour mesurer la normalité de la démarche., The problem of assessing human gaits has received a great attention in the literature since gait analysis is one of key components in healthcare. Marker-based and multi-camera systems are widely employed to deal with this problem. However, such systems usually require specific equipments with high price and/or high computational cost. In order to reduce the cost of devices, we focus on a system of gait analysis which employs only one depth sensor. The principle of our work is similar to multi-camera systems, but the collection of cameras is replaced by one depth sensor and mirrors. Each mirror in our setup plays the role of a camera which captures the scene at a different viewpoint. Since we use only one camera, the step of synchronization can thus be avoided and the cost of devices is also reduced. Our studies can be separated into two categories: 3D reconstruction and gait analysis. The result of the former category is used as the input of the latter one. Our system for 3D reconstruction is built with a depth camera and two mirrors. Two types of depth sensor, which are distinguished based on the scheme of depth estimation, have been employed in our works. With the structured light (SL) technique integrated into the Kinect 1, we perform the 3D reconstruction based on geometrical optics. In order to increase the level of details of the 3D reconstructed model, the Kinect 2 with time-of-flight (ToF) depth measurement is used for image acquisition instead of the previous generation. However, due to multiple reflections on the mirrors, depth distortion occurs in our setup. We thus propose a simple approach for reducing such distortion before applying geometrical optics to reconstruct a point cloud of the 3D object. For the task of gait analysis, we propose various alternative approaches focusing on the problem of gait normality/symmetry measurement. They are expected to be useful for clinical treatments such as monitoring patient's recovery after surgery. These methods consist of model-free and model-based approaches that have different cons and pros. In this dissertation, we present 3 methods that directly process point clouds reconstructed from the previous work. The first one uses cross-correlation of left and right half-bodies to assess gait symmetry while the other ones employ deep auto-encoders to measure gait normality.

В статье на основе анализа более 400 судебных решений проведено исследование доказывания по уголовным делам частного обвинения. Рассмотрены позиции судов различного уровня, на примере которых продемонстрировано несовершенство регламентации указанной процедуры в УПК РФ, что приводит к ущемлению закрепленных в Конституции РФ прав и свобод участников уголовного судопроизводства, а также осложняет деятельность как мировых судей, так и судей проверочной инстанции. Продемонстрирована актуальность исследований процесса доказывания по уголовным делам указанной категории, поскольку последние могут быть использованы как средство решения хозяйственных споров, в связи с чем поставлен вопрос о необходимости усиления действия принципа состязательности путем расширения полномочий сторон по сбору и представлению суду доказательств как средства обеспечения стабильности хозяйственного оборота и воспрепятствования незаконной конкуренции путем использования мер уголовно-процессуального воздействия по делам частного обвинения. Выявлены и проанализированы особенности реализации состязательности сторон при рассмотрении уголовных дел частного обвинения. Проведен анализ существующей практики сбора доказательств сторонами и судом, оценки полученных доказательств судом и их интерпретация при вынесении итогового решения на основании внутреннего убеждения. Дана оценка деятельности суда апелляционной инстанции по проверке приговоров мировых судов и их подход к оценке доказательств. Приведены примеры судебных решений и проанализирована их внутренняя логика, указано на недостаточное следование положениям о презумпции невиновности и доказанности вины. Сформулирован вывод о необходимости усиления состязательных начал в отправлении правосудия путем внесения изменений в ч. 3 ст. 86 УПК РФ, he article based on the analysis of more than 400 court decisions presents the study of proof in criminal cases of private charge. The positions of courts of different levels are examined, which demonstrated imperfection of regulation of this procedure in the code of criminal procedure, which leads to the denial enshrined in the Constitution of the Russian Federation of the rights and freedoms of participants in criminal proceedings, and also complicates the activities of magistrates and judges of the validation court. The relevance of research of the process of proof in criminal cases of this category is demonstrated, since the latter can be used as a means of resolving economic disputes, in connection with which the question is raised about the need to strengthen the operation of the adversarial principle by expanding the powers of the parties to collect and present evidence to the court as a means of ensuring the stability of economic turnover and preventing illegal competition by using measures of criminal procedural influence in the cases of private prosecution. Features of realization of competitiveness of the parties at consideration of criminal cases of private charge are revealed and analyzed. The analysis of the existing practice of collecting evidence by the parties and the court is provided. The process of evaluation of the evidences obtained by the court and their interpretation in making the final decision is analyzed. The specificity of the evaluation of the evidence base by the world court on the basis of internal conviction is considered. The assessment of activity of the court of appeal on checking the sentences of the world courts and their approach to assessment of the proofs is given. Examples of the court decisions are given and their internal logic is analyzed, insufficient adherence to the provisions on the presumption of innocence and proof of guilt is indicated. The conclusion is formulated about the need to strengthen adversarial principles in the administration of justice by amending part 3 of article 86 of the criminal procedure code., №1(50) (2020)

Artificial Intelligence (AI) is one of the hottest topics in our society, especially when it comes to solving data-analysis problems. Industry are conducting their digital shifts, and AI is becoming a cornerstone technology for making decisions out of the huge amount of (sensors-based) data available in the production floor. However, such technology may be disappointing when deployed in real conditions. Despite good theoretical performances and high accuracy when trained and tested in isolation, a Machine-Learning (M-L) model may provide degraded performances in real conditions. One reason may be fragility in treating properly unexpected or perturbed data. The objective of the paper is therefore to study the robustness of seven M-L and Deep-Learning (D-L) algorithms, when classifying univariate time-series under perturbations. A systematic approach is proposed for artificially injecting perturbations in the data and for evaluating the robustness of the models. This approach focuses on two perturbations that are likely to happen during data collection. Our experimental study, conducted on twenty sensors’ datasets from the public University of California Riverside (UCR) repository, shows a great disparity of the models’ robustness under data quality degradation. Those results are used to analyse whether the impact of such robustness can be predictable—thanks to decision trees—which would prevent us from testing all perturbations scenarios. Our study shows that building such a predictor is not straightforward and suggests that such a systematic approach needs to be used for evaluating AI models’ robustness.

Adversarial patch attacks are among one of the most practical threat models against real-world computer vision systems. This paper studies certified and empirical defenses against patch attacks. We begin with a set of experiments showing that most existing defenses, which work by pre-processing input images to mitigate adversarial patches, are easily broken by simple white-box adversaries. Motivated by this finding, we propose the first certified defense against patch attacks, and propose faster methods for its training. Furthermore, we experiment with different patch shapes for testing, obtaining surprisingly good robustness transfer across shapes, and present preliminary results on certified defense against sparse attacks. Our complete implementation can be found on: https://github.com/Ping-C/certifiedpatchdefense., International Conference on Learning Representations, ICLR 2020

With the method of conversation analysis, this study examines the level of adverseness in questions between members of parliament from different parties. The data consists of question – response sequences derived from a ministerial statement from the prime minister in the House of Commons. This study finds that, in question – response sequences between oppositional members of parliament and the prime minister, adversarial presuppositions in questions can be used as a strategy to project negative traits upon the respondent. Adversarial dimensions of hostility, assertiveness and directness can also be found in adversarial questions. In these instances, the respondent may adjust their answer to match the level of adverseness from the questioner through the use of certain lexis, creating counter sequences. Adversarial questions are the most common type of question from members of the oppositional party, and there are different adversarial strategies being used. Questions from members of the government party do not make use of adversarial strategies, and should not be described as adversarial.

ransfer learning, in which a network is trained on one task and re-purposed on another, is often used to produce neural network classifiers when data is scarce or full-scale training is too costly. When the goal is to produce a model that is not only accurate but also adversarially robust, data scarcity and computational limitations become even more cumbersome. We consider robust transfer learning, in which we transfer not only performance but also robustness from a source model to a target domain. We start by observing that robust networks contain robust feature extractors. By training classifiers on top of these feature extractors, we produce new models that inherit the robustness of their parent networks. We then consider the case of "fine tuning" a network by re-training end-to-end in the target domain. When using lifelong learning strategies, this process preserves the robustness of the source network while achieving high accuracy. By using such strategies, it is possible to produce accurate and robust models with little data, and without the cost of adversarial training. Additionally, we can improve the generalization of adversarially trained models, while maintaining their robustness.

Рассматриваются отдельные аспекты состязательности в уголовном судопроизводстве как основополагающего принципа уголовного процесса, участия сторон в доказывании и оценки доказательств, наделении их определенными полномочиями на досудебной стадии уголовного процесса и законодательном закреплении этих полномочий, обеспечении охраны прав и законных интересов участников уголовного судопроизводства. Анализируется возможность законодательного совершенствования состязательности на досудебной стадии уголовного судопроизводства, освещается характер взаимоотношений обвинения и защиты при производстве расследования уголовного дела., Some aspects of the principle of competition in criminal proceedings, as a fundamental principle of the criminal process, participation of the parties in proving and evaluating evidence, granting them certain powers at the pre-trial stage of the criminal process and legislative consolidation of these powers, ensuring the protection of the rights and legitimate interests of participants in criminal proceedings areconsidered. The article analyzes the possibility of legislative improvement of competition at the pre-trial stage of criminal proceedings, the nature of the relationship between the prosecution and the defense during the investigation of a criminal case.

With the method of conversation analysis, this study examines the level of adverseness in questions between members of parliament from different parties. The data consists of question – response sequences derived from a ministerial statement from the prime minister in the House of Commons. This study finds that, in question – response sequences between oppositional members of parliament and the prime minister, adversarial presuppositions in questions can be used as a strategy to project negative traits upon the respondent. Adversarial dimensions of hostility, assertiveness and directness can also be found in adversarial questions. In these instances, the respondent may adjust their answer to match the level of adverseness from the questioner through the use of certain lexis, creating counter sequences. Adversarial questions are the most common type of question from members of the oppositional party, and there are different adversarial strategies being used. Questions from members of the government party do not make use of adversarial strategies, and should not be described as adversarial.

Open-source software is open to anyone by design, whether it is a community of developers, hackers or malicious users. Authors of open-source software typically hide their identity through nicknames and avatars. However, they have no protection against authorship attribution techniques that are able to create software author profiles just by analyzing software characteristics. In this paper we present an author imitation attack that allows to deceive current authorship attribution systems and mimic a coding style of a target developer. Withing this context we explore the potential of the existing attribution techniques to be deceived. Our results show that we are able to imitate the coding style of the developers based on the data collected from the popular source code repository, GitHub. To subvert author imitation attack, we propose a novel author obfuscation approach that allows us to hide the coding style of the author. Unlike existing obfuscation tools, this new obfuscation technique uses transformations that preserve code readability. We assess the effectiveness of our attacks on several datasets produced by actual developers from GitHub, and participants of the GoogleCodeJam competition. Throughout our experiments we show that the author hiding can be achieved by making sensible transformations which significantly reduce the likelihood of identifying the author's style to 0% by current authorship attribution systems.

Esta investigación analiza, critica y propone los límites que debe tener un interviniente en el proceso penal Colombiano, es decir, la necesidad de la presencia o no del Procurador Judicial y la Victima en la etapa de juzgamiento dentro de un sistema con tendencia acusatoria, el cual se caracteriza por ser de carácter adversarial. Por ello con la presencia de estos dos sujetos procesales se originan una serie de problemáticas que se analizaran en este trabajo de grado. Para llegar abordar el tema de los límites de estos sujetos procesales, primero se analizará los sistemas procesales penales, las partes de un sistema acusatorio, es decir, la Fiscalía y la Defensa, las funciones del Procurador Judicial y la Victima en el proceso penal Colombiano, las problemáticas que se generan con esta desnaturalización de un sistema adversarial y por último las posturas que se adoptaron en esta investigación, teniendo en cuenta la Ley, la Jurisprudencia de las Altas Cortes y la Doctrina. This research analyzes, criticizes and proposes the limits that an intervener must have in the Colombian criminal process, that is, the need for the presence or absence of the Judicial Procurator and the Victim in the judging stage within a system with an accusatory tendency, the which is characterized by being adversarial. Therefore, with the presence of these two procedural subjects originate a series of problems that will be analyzed in this degree work. In order to deal with the issue of the limits of these procedural subjects, first we will analyze the criminal procedural systems, the parts of an accusatory system, that is, the Office of the Prosecutor and the Defense, the functions of the Judicial Procurator and the Victim in the Colombian criminal process. , the problems that are generated with this denaturing of an adversarial system and finally the positions adopted in this investigation, taking into account the Law, the Jurisprudence of the High Courts and the Doctrine. Abogado http://www.ustavillavicencio.edu.co/home/index.php/unidades/extension-y-proyeccion/investigacion Pregrado