Search

In this paper, we propose a novel adversarial defence mechanism for image classification -- CARSO -- inspired by cues from cognitive neuroscience. The method is synergistically complementary to adversarial training and relies on knowledge of the internal representation of the attacked classifier. Exploiting a generative model for adversarial purification, conditioned on such representation, it samples reconstructions of inputs to be finally classified. Experimental evaluation by a well-established benchmark of varied, strong adaptive attacks, across diverse image datasets and classifier architectures, shows that CARSO is able to defend the classifier significantly better than state-of-the-art adversarial training alone -- with a tolerable clean accuracy toll. Furthermore, the defensive architecture succeeds in effectively shielding itself from unforeseen threats, and end-to-end attacks adapted to fool stochastic defences. Code and pre-trained models are available at https://github.com/emaballarin/CARSO ., 20 pages, 5 figures, 10 tables; Update: removed visual artifacts from some figures, fixed typos/capitalisation, typographic/pagination improvements

This document presents the mission and strategy of the Italian Competence Centre on Open Science, FAIR, and EOSC. The Competence Centre is an initiative born within the Italian Computing and Data Infrastructure (ICDI), a forum created by representatives of major Italian Research Infrastructures and e-Infrastructures, with the aim of promoting sinergies at the national level, and optimising the Italian participation to European and global challenges in this field, including the European Open Science Cloud (EOSC), the European Data Infrastructure (EDI) and HPC. This working paper depicts the mission and objectives of the ICDI Competence Centre, a network of experts with various skills and competences that are supporting the national stakeholders on topics related to Open Science, FAIR principles application and participation to the EOSC. The different actors and roles are described in the document as well as the activities and services offered, and the added value each stakeholder can find the in Competence Centre. The tools and services provided, in particular the concept for the portal, though which the Centre will connect to the national landscape and users, are also presented. This record is the english translation of the original Italian (2021). Competence Centre ICDI per Open Science, FAIR, ed EOSC - Mission, Strategia e piano d'azione. Zenodo. https://doi.org/10.5281/zenodo.5071055, This record is the english translation of the original Italian (2021). Competence Centre ICDI per Open Science, FAIR, ed EOSC - Mission, Strategia e piano d'azione. Zenodo. https://doi.org/10.5281/zenodo.5071055

This document presents the mission and strategy of the Italian Competence Centre on Open Science, FAIR, and EOSC. The Competence Centre is an initiative born within the Italian Computing and Data Infrastructure (ICDI), a forumcreated by representatives of major Italian Research Infrastructures and e-Infrastructures, with the aim of promoting sinergies at the national level, and optimising the Italian participation to European and global challenges in this field, including the European Open Science Cloud (EOSC), the European Data Infrastructure (EDI) andHPC. This working paper depicts the mission and objectives of the ICDI Competence Centre, a network of experts with various skills and competences that are supporting the national stakeholders on topics related to Open Science, FAIR principles application and participation to the EOSC. The different actors and roles are described in the document as well as the activities and services offered, and the added value each stakeholder can find the in Competence Centre. The tools and services provided, in particular the concept for the portal, though which the Centre will connect to the national landscape and users, are also presented. An english translation of this document is provided in Zenodo:Lazzeri, Emma, et all. (2021). ICDI Competence Centre for Open Science, FAIR and EOSC - Mission, Strategy and Action Plan. Zenodo. https://doi.org/10.5281/zenodo.5512638&nbsp

In recent years, Artificial Neural Networks (ANNs) pruning has become the focal point of many researches, due to the extreme overparametrization of such models. This has urged the scientific world to investigate methods for the simplification of the structure of weights in ANNs, mainly in an effort to reduce time for both training and inference. Frankle and Carbin [1], and later Renda, Frankle, and Carbin [2] introduced and refined an iterative pruning method which is able to effectively prune the network of a great portion of its parameters with little to no loss in performance. On the downside, this method requires a large amount of time for its application, since, for each iteration, the network has to be trained for (almost) the same amount of epochs of the unpruned network. In this work, we show that, for a limited setting, if targeting high overall sparsity rates, this time can be effectively reduced for each iteration, save for the last one, by more than 50%, while yielding a final product (i.e., final pruned network) whose performance is comparable to the ANN obtained using the existing method.

Simple Summary Dyskerin is a nucleolar protein involved in the modification and processing of ribosomal RNA and in the stabilization of the telomerase RNA component. In several human tumors, including breast cancer, dyskerin overexpression is found related to patients’ worse prognosis. Our aim was to study this phenomenon at the molecular and cellular levels. We firstly confirmed the correlation between high dyskerin expression with patients’ shorter survival. Then, through the generation of cellular models of increased dyskerin expression, we found that increasing dyskerin levels conferred a more aggressive phenotype and increased intrinsic ribosomal activity only in cells derived from normal breast epithelium. Our study provides evidence on the prognostic and bio-pathological relevance of the overexpression of dyskerin in breast carcinoma. A possible mechanistic explanation of the effects of dyskerin overexpression, involving specific ribosomal RNA modification and consequent increased ribosomal activity, is also provided. Abstract Dyskerin is a nucleolar protein involved in the small nucleolar RNA (snoRNA)-guided pseudouridylation of specific uridines on ribosomal RNA (rRNA), and in the stabilization of the telomerase RNA component (hTR). Loss of function mutations in DKC1 causes X-linked dyskeratosis congenita, which is characterized by a failure of proliferating tissues and increased susceptibility to cancer. However, several tumors show dyskerin overexpression. We observed that patients with primary breast cancers with high dyskerin levels are more frequently characterized by shorter survival rates and positive lymph node status than those with tumors with a lower dyskerin expression. To functionally characterize the effects of high dyskerin expression, we generated stably overexpressing DKC1 models finding that increased dyskerin levels conferred a more aggressive cellular phenotype in untransformed immortalized MCF10A cells. Contextually, DKC1 overexpression led to an upregulation of some snoRNAs, including SNORA67 and a significantly increased U1445 modification on 18S rRNA, the known target of SNORA67. Lastly, we found that dyskerin overexpression strongly enhanced the synthetic activity of ribosomes increasing translational efficiency in MCF10A. Altogether, our results indicate that dyskerin may sustain the neoplastic phenotype from an early stage in breast cancer endowing ribosomes with an augmented translation efficiency.

Deep Neural Networks (DNNs) are essential tools of modern science and technology. The current lack of explainability of their inner workings and of principled ways to tame their architectural complexity triggered a lot of research in recent years. There is hope that, by making sense of representations in their hidden layers, we could collect insights on how to reduce model complexity—without performance degradation—by pruning useless connections. It is natural then to ask the following question: how similar are representations in pruned and unpruned models? Even small insights could help in finding principled ways to design good lightweight models, enabling significant savings of computation, memory, time and energy. In this work, we investigate empirically this problem on a wide spectrum of similarity measures, network architectures and datasets. We find that the results depend critically on the similarity measure used and we discuss briefly the origin of these differences, concluding that further investigations are required in order to make substantial advances.

Three-dimensional carbon nanotube frameworks favor spinal cord explant rewiring of motor outputs., In modern neuroscience, significant progress in developing structural scaffolds integrated with the brain is provided by the increasing use of nanomaterials. We show that a multiwalled carbon nanotube self-standing framework, consisting of a three-dimensional (3D) mesh of interconnected, conductive, pure carbon nanotubes, can guide the formation of neural webs in vitro where the spontaneous regrowth of neurite bundles is molded into a dense random net. This morphology of the fiber regrowth shaped by the 3D structure supports the successful reconnection of segregated spinal cord segments. We further observed in vivo the adaptability of these 3D devices in a healthy physiological environment. Our study shows that 3D artificial scaffolds may drive local rewiring in vitro and hold great potential for the development of future in vivo interfaces.

During the last few decades, artificial neural networks (ANN) have achieved an enormous success in regression and classification tasks. The empirical success has not been matched with an equally strong theoretical understanding of such models, as some of their working principles (training dynamics, generalization properties, and the structure of inner representations) still remain largely unknown. It is, for example, particularly difficult to reconcile the well known fact that ANNs achieve remarkable levels of generalization also in conditions of severe over-parametrization. In our work, we explore a recent network compression technique, called Iterative Magnitude Pruning (IMP), and apply it to convolutional neural networks (CNN). The pruned and unpruned models are compared layer-wise with Canonical Correlation Analysis (CCA). Our results show a high similarity between layers of pruned and unpruned CNNs in the first convolutional layers and in the fully-connected layer, while for the intermediate convolutional layers the similarity is significantly lower. This suggests that, although in intermediate layers representation in pruned and unpruned networks is markedly different, in the last part the fully-connected layers act as pivots, producing not only similar performances but also similar representations of the data, despite the large difference in the number of parameters involved.