Search

The application for random numbers is ubiquitous. We experimentally build a well-studied quantum random number generator from homodyne measurements on the quadrature of the vacuum fluctuations. Semi-device-independence in this random number generator is usually obtained using phase modulators to shift the phase of the laser and obtain random sampling from both X and P quadrature measurements of the vacuum state in previous implementations. We characterize the experimental parameters for optimal performance of this source-device independent quantum random number generator by measuring the two quadratures concurrently using two homodyne detectors. We also study the influence of these parameters on randomness, which can be extracted based on Shannon entropy and von Neumann entropy, which correspond to an eavesdropper listening to classical and quantum side information, respectively.

Harnessing pre-trained LLMs to improve ASR systems, particularly for low-resource languages, is now an emerging area of research. Existing methods range from using LLMs for ASR error correction to tightly coupled systems that replace the ASR decoder with the LLM. These approaches either increase decoding time or require expensive training of the cross-attention layers. We propose SALSA, which couples the decoder layers of the ASR to the LLM decoder, while synchronously advancing both decoders. Such coupling is performed with a simple projection of the last decoder state, and is thus significantly more training efficient than earlier approaches. A challenge of our proposed coupling is handling the mismatch between the tokenizers of the LLM and ASR systems. We handle this mismatch using cascading tokenization with respect to the LLM and ASR vocabularies. We evaluate SALSA on 8 low-resource languages in the FLEURS benchmark, yielding substantial WER reductions of up to 38%., Comment: Accepted to INTERSPEECH 2024

This work contributes towards the development of an efficient and scalable open-source Secure Multi-Party Computation (SMPC) protocol on machines with moderate computational resources. We use the ABY2.0 SMPC protocol implemented on the C++ based MOTION2NX framework for secure convolutional neural network (CNN) inference application with semi-honest security. Our list of contributions are as follows. Firstly, we enhance MOTION2NX by providing a tensorized version of several primitive functions including the Hadamard product, indicator function and argmax function. Our design of secure indicator function based on a novel approach that uses secure Relu function available in the baseline MOTION2NX implementation. The secure indicator function is used, in turn, as a building block for a novel implementation of secure argmax. Secondly, we also develop a novel splitting of the computations at each CNN layer into multiple configurable chunks thereby resulting in significant reduction in RAM usage. Thirdly, we adapt an existing Helper node algorithm, working in tandem with the ABY2.0 protocol, for efficient convolution computation. This algorithm not only reduces execution time but also reduces the RAM usage required to execute CNN models, but comes at a cost of an additional compute server. Moreover, the ideas presented in this paper can also be applied to secure neural network training., Comment: 20 pages, 1 figure. arXiv admin note: text overlap with arXiv:2310.10133

There exists three main areas of study inside of the field of predictive maintenance: anomaly detection, fault diagnosis, and remaining useful life prediction. Notably, anomaly detection alerts the stakeholder that an anomaly is occurring. This raises two fundamental questions: what is causing the fault and how can we fix it? Inside of the field of explainable artificial intelligence, counterfactual explanations can give that information in the form of what changes to make to put the data point into the opposing class, in this case "healthy". The suggestions are not always actionable which may raise the interest in asking "what if we do this instead?" In this work, we provide a proof of concept for utilizing counterfactual explanations as what-if analysis. We perform this on the PRONOSTIA dataset with a temporal convolutional network as the anomaly detector. Our method presents the counterfactuals in the form of a what-if analysis for this base problem to inspire future work for more complex systems and scenarios., Comment: 8 pages, 6 figures, 3 tables

Qualitative data collection and analysis approaches, such as those employing interviews and focus groups, provide rich insights into customer attitudes, sentiment, and behavior. However, manually analyzing qualitative data requires extensive time and effort to identify relevant topics and thematic insights. This study proposes a novel approach to address this challenge by leveraging Retrieval Augmented Generation (RAG) based Large Language Models (LLMs) for analyzing interview transcripts. The novelty of this work lies in strategizing the research inquiry as one that is augmented by an LLM that serves as a novice research assistant. This research explores the mental model of LLMs to serve as novice qualitative research assistants for researchers in the talent management space. A RAG-based LLM approach is extended to enable topic modeling of semi-structured interview data, showcasing the versatility of these models beyond their traditional use in information retrieval and search. Our findings demonstrate that the LLM-augmented RAG approach can successfully extract topics of interest, with significant coverage compared to manually generated topics from the same dataset. This establishes the viability of employing LLMs as novice qualitative research assistants. Additionally, the study recommends that researchers leveraging such models lean heavily on quality criteria used in traditional qualitative research to ensure rigor and trustworthiness of their approach. Finally, the paper presents key recommendations for industry practitioners seeking to reconcile the use of LLMs with established qualitative research paradigms, providing a roadmap for the effective integration of these powerful, albeit novice, AI tools in the analysis of qualitative datasets within talent, Comment: Accepted to KDD '24 workshop on Talent Management and Computing (TMC 2024). 9 pages

In image processing, one of the most challenging tasks is to render an image's semantic meaning using a variety of artistic approaches. Existing techniques for arbitrary style transfer (AST) frequently experience mode-collapse, over-stylization, or under-stylization due to a disparity between the style and content images. We propose a novel framework called D$^2$Styler (Discrete Diffusion Styler) that leverages the discrete representational capability of VQ-GANs and the advantages of discrete diffusion, including stable training and avoidance of mode collapse. Our method uses Adaptive Instance Normalization (AdaIN) features as a context guide for the reverse diffusion process. This makes it easy to move features from the style image to the content image without bias. The proposed method substantially enhances the visual quality of style-transferred images, allowing the combination of content and style in a visually appealing manner. We take style images from the WikiArt dataset and content images from the COCO dataset. Experimental results demonstrate that D$^2$Styler produces high-quality style-transferred images and outperforms twelve existing methods on nearly all the metrics. The qualitative results and ablation studies provide further insights into the efficacy of our technique. The code is available at https://github.com/Onkarsus13/D2Styler., Comment: Paper accepted at 27th International Conference on Pattern Recognition (ICPR), 2024

Modern data centres are increasingly adopting containers to enhance power and performance efficiency. These data centres consist of multiple heterogeneous machines, each equipped with varying amounts of resources such as CPU, I/O, memory, and network bandwidth. Data centers rent their resources to applications, which demand different amounts of resources and execute on machines for extended durations if the machines provide the demanded resources to the applications. Certain applications run efficiently on specific machines, referred to as system affinity between applications and machines. In contrast, others are incompatible with specific machines, referred to as anti-affinity between applications and machines. We consider that there are multiple applications, and data centers need to execute as many applications as possible. Data centers incur electricity based on CPU usage due to the execution of applications, with the cost being proportional to the cube of the total CPU usage. It is a challenging problem to place applications on the machines they have an affinity for while keeping the electricity cost in check. Our work addresses the placement problem of matching applications to machines to minimize overall electricity costs while maximizing the number of affinity pairs of machines and applications. We propose three solution approaches: (a) Power-Aware Placement (PAP): applications are placed on machines where power usage is minimized, (b) Affinity-Aware Placement (AAP): applications are placed on machines where affinity is maximized, (c) Combined Power-Affinity Placement (CPAAP): this approach integrates the benefits of both PAP and AAP. Our proposed approach improves the affinity satisfaction ratio by up to 4% while reducing the total system cost by up to 26% and improving the affinity payoff ratio by up to 37% compared to state-of-the-art approaches for real-life datasets.

This research investigates biases in text-to-image (TTI) models for the Indic languages widely spoken across India. It evaluates and compares the generative performance and cultural relevance of leading TTI models in these languages against their performance in English. Using the proposed IndicTTI benchmark, we comprehensively assess the performance of 30 Indic languages with two open-source diffusion models and two commercial generation APIs. The primary objective of this benchmark is to evaluate the support for Indic languages in these models and identify areas needing improvement. Given the linguistic diversity of 30 languages spoken by over 1.4 billion people, this benchmark aims to provide a detailed and insightful analysis of TTI models' effectiveness within the Indic linguistic landscape. The data and code for the IndicTTI benchmark can be accessed at https://iab-rubric.org/resources/other-databases/indictti., Comment: Accepted in ECCV 2024

Modeling and Simulation (M&S) is finding increasing application in development and testing of command and control systems comprised of information-intensive component systems. Achieving interoperability is one of the chief System of Systems (SoS) engineering objectives in the development of command and control (C2) capabilities for joint and coalition warfare. In this paper, we apply an SoS perspective on the integration of M&S with such systems. We employ recently developed interoperability concepts based on linguistic categories along with the Discrete Event System Specification (DEVS) formalism to implement a standard for interoperability. We will show how the developed standard is implemented in DEVS/SOA net-centric modeling and simulation framework that uses XML-based Service Oriented Architecture (SOA). We will discuss the simulator interfaces and the design issues in their implementation in DEVS/SOA. We will illustrate the application of DEVS/SOA in a multi-agent test instrumentation system that is deployable as a SOA., Comment: arXiv admin note: substantial text overlap with arXiv:2407.03686

Parrondo's paradox, a counterintuitive phenomenon where two losing strategies combine to produce a winning outcome, has been a subject of interest across various scientific fields, including quantum mechanics. In this study, we investigate the manifestation of Parrondo's paradox in discrete-time quantum walks. We demonstrate the existence of Parrondo's paradox using space and time-dependent coins without the need for a higher-dimensional coin or adding decoherence to the system. Our results enhance the feasibility of practical implementations and provide deeper insights into the underlying quantum dynamics, specifically the propagation constrained by the interference pattern of quantum walks. The implications of our results suggest the potential for more accessible and efficient designs in quantum transport, broadening the scope and application of Parrondo's paradox beyond conventional frameworks., Comment: Comments and suggestions are most welcome

Earth went through at least two periods of global glaciation (i.e., ``Snowball Earth'' states) during the Neoproterozoic, the shortest of which (the Marinoan) may not have lasted sufficiently long for its termination to be explained by the gradual volcanic build-up of greenhouse gases in the atmosphere. Large asteroid impacts and supervolcanic eruptions have been suggested as stochastic geological events that could cause a sudden end to global glaciation via a runaway melting process. Here, we employ an energy balance climate model to simulate the evolution of Snowball Earth's surface temperature after such events. We find that even a large impactor (diameters of $d \sim 100\,\mathrm{km}$) and the supervolcanic Toba eruption ($74\,\mathrm{kyr}$ ago), are insufficient to terminate a Snowball state unless background CO$_2$ has already been driven to high levels by long-term outgassing. We suggest, according to our modeling framework, that Earth's Snowball states would have been resilient to termination by stochastic events., Comment: Accepted for publication in Geophysical Research Letters (GRL)

Determining the approximate degree composition for Boolean functions remains a significant unsolved problem in Boolean function complexity. In recent decades, researchers have concentrated on proving that approximate degree composes for special types of inner and outer functions. An important and extensively studied class of functions are the recursive functions, i.e.~functions obtained by composing a base function with itself a number of times. Let $h^d$ denote the standard $d$-fold composition of the base function $h$. The main result of this work is to show that the approximate degree composes if either of the following conditions holds: \begin{itemize} \item The outer function $f:\{0,1\}^n\to \{0,1\}$ is a recursive function of the form $h^d$, with $h$ being any base function and $d= \Omega(\log\log n)$. \item The inner function is a recursive function of the form $h^d$, with $h$ being any constant arity base function (other than AND and OR) and $d= \Omega(\log\log n)$, where $n$ is the arity of the outer function. \end{itemize} In terms of proof techniques, we first observe that the lower bound for composition can be obtained by introducing majority in between the inner and the outer functions. We then show that majority can be \emph{efficiently eliminated} if the inner or outer function is a recursive function.

Modeling and Simulation (M&S) for system design and prototyping is practiced today both in the industry and academia. M&S are two different areas altogether and have specific objectives. However, most of the times these two separate areas are taken together. The developed code is tightly woven around both the model and the underlying simulator that executes it. This constraints both the model development and the simulation engine that impacts scalability of the developed code. Furthermore, a lot of time is spent in development of a model because it needs both domain knowledge and simulation techniques, which also requires communication among users and developers. Unified Modeling Language (UML) is widely accepted in the industry, whereas Discrete Event Specification (DEVS) based modeling that separates the model and the simulator, provides a cleaner methodology to develop models and is much used in academia. DEVS today is used by engineers who understand discrete event modeling at a much detailed level and are able to translate requirements to DEVS modeling code. There have been earlier efforts to integrate UML and DEVS but they haven't succeeded in providing a transformation mechanism due to inherent differences in these two modeling paradigms. This paper presents an integrated approach towards crosstransformations between UML and DEVS using the proposed eUDEVS, which stands for executable UML based on DEVS. Further, we will also show that the obtained DEVS models belong to a specific class of DEVS models called Finite Deterministic DEVS (FD-DEVS) that is available as a W3C XML Schema in XFD-DEVS. We also put the proposed eUDEVS in a much larger unifying framework called DEVS Unified Process that allows bifurcated model-continuity based lifecycle methodology for systems M&S. Finally, we demonstrate the laid concepts with a complete example.

Discrete EVent Specification (DEVS) environments are known to be implemented over middleware systems such as HLA, RMI, CORBA and others. DEVS exhibits concepts of systems theory and modeling and supports capturing the system behavior from the physical and behavioral perspectives. Further, they are implemented using Object-oriented languages like Java and C++. This research work uses the Java platform to implement DEVS over a Service Oriented Architecture (SOA) framework. Called the DEVS/SOA, the framework supports a development and testing environment known as DEVS Unified Process that is built on a model-continuity-based life cycle methodology. DEVS Unified Process allows DEVS-based Modeling and Simulation (M&S) over net-centric platforms using DEVS/SOA. This framework also provides the crucial feature of run-time composability of coupled systems using SOA. We describe the architecture and designs of the both the server and the client. The client application communicates with multiple servers hosting DEVS simulation services. These Simulation services are developed using the proposed symmetrical services architecture wherein the server can act as both a service provider and a service consumer contrary to the unidirectional client-server paradigm. We also discuss how this Services based architecture provides solutions for cross-platform distributed M&S. We demonstrate DEVS/SOA framework with a scenario of Joint Close Air Support specified in Business Process Modeling Notation (BPMN). We also provide a real-world application of Network health monitoring using DEVS/SOA layered architectural framework.

Large Language Models (LLMs) have showcased impressive abilities in generating fluent responses to diverse user queries. However, concerns regarding the potential misuse of such texts in journalism, educational, and academic contexts have surfaced. SemEval 2024 introduces the task of Multigenerator, Multidomain, and Multilingual Black-Box Machine-Generated Text Detection, aiming to develop automated systems for identifying machine-generated text and detecting potential misuse. In this paper, we i) propose a RoBERTa-BiLSTM based classifier designed to classify text into two categories: AI-generated or human ii) conduct a comparative study of our model with baseline approaches to evaluate its effectiveness. This paper contributes to the advancement of automatic text detection systems in addressing the challenges posed by machine-generated text misuse. Our architecture ranked 46th on the official leaderboard with an accuracy of 80.83 among 125., Comment: SemEval-2024

Industry 4.0 has witnessed the rise of complex robots fueled by the integration of Artificial Intelligence/Machine Learning (AI/ML) and Digital Twin (DT) technologies. While these technologies offer numerous benefits, they also introduce potential privacy and security risks. This paper surveys privacy attacks targeting robots enabled by AI and DT models. Exfiltration and data leakage of ML models are discussed in addition to the potential extraction of models derived from first-principles (e.g., physics-based). We also discuss design considerations with DT-integrated robotics touching on the impact of ML model training, responsible AI and DT safeguards, data governance and ethical considerations on the effectiveness of these attacks. We advocate for a trusted autonomy approach, emphasizing the need to combine robotics, AI, and DT technologies with robust ethical frameworks and trustworthiness principles for secure and reliable AI robotic systems., Comment: 10 pages, 3 figures, 1 table

Contribution of resolved and unresolved extragalactic point sources to the low-frequency sky spectrum is a potentially non-negligible part of the astrophysical foregrounds for cosmic dawn 21-cm experiments. The clustering of such point sources on the sky, combined with the frequency-dependence of the antenna beam, can also make this contribution chromatic. By combining low-frequency measurements of the luminosity function and the angular correlation function of extragalactic point sources, we develop a model for the contribution of these sources to the low-frequency sky spectrum. Using this model, we find that the contribution of sources with flux density $>10^{-6}\,$Jy to the sky-averaged spectrum is smooth and of the order of a few kelvins at 50 - $200\,$MHz. We combine this model with measurements of the galactic foreground spectrum and convolve the result with the beam of the conical log-spiral antenna planned as part of the Radio Experiment for the Analysis of Cosmic Hydrogen (REACH) project. We find that the contribution of point sources to resultant spectrum is $\sim0.4\%$ of the total foregrounds, but still larger by at least an order of magnitude than the standard predictions for the cosmological 21-cm signal. As a result, not accounting for the point-source contribution leads to a systematic bias in 21-cm signal recovery. We show, however, that in the REACH case, this reconstruction bias can be removed by modelling the point-source contribution as a power law with a running spectral index. We make our code publicly available as a Python package labelled epspy., Comment: 13 pages, 14 figures, and 1 appendix. Submitted to MNRAS. Comments are welcome

Existing methods for adapting large language models (LLMs) to new tasks are not suited to multi-task adaptation because they modify all the model weights -- causing destructive interference between tasks. The resulting effects, such as catastrophic forgetting of earlier tasks, make it challenging to obtain good performance on multiple tasks at the same time. To mitigate this, we propose Lottery Ticket Adaptation (LoTA), a sparse adaptation method that identifies and optimizes only a sparse subnetwork of the model. We evaluate LoTA on a wide range of challenging tasks such as instruction following, reasoning, math, and summarization. LoTA obtains better performance than full fine-tuning and low-rank adaptation (LoRA), and maintains good performance even after training on other tasks -- thus, avoiding catastrophic forgetting. By extracting and fine-tuning over lottery tickets (or sparse task vectors), LoTA also enables model merging over highly dissimilar tasks. Our code is made publicly available at https://github.com/kiddyboots216/lottery-ticket-adaptation.

The electricity grid has evolved from a physical system to a cyber-physical system with digital devices that perform measurement, control, communication, computation, and actuation. The increased penetration of distributed energy resources (DERs) that include renewable generation, flexible loads, and storage provides extraordinary opportunities for improvements in efficiency and sustainability. However, they can introduce new vulnerabilities in the form of cyberattacks, which can cause significant challenges in ensuring grid resilience. %, i.e. the ability to rapidly restore grid services in the face of severe disruptions. We propose a framework in this paper for achieving grid resilience through suitably coordinated assets including a network of Internet of Things (IoT) devices. A local electricity market is proposed to identify trustable assets and carry out this coordination. Situational Awareness (SA) of locally available DERs with the ability to inject power or reduce consumption is enabled by the market, together with a monitoring procedure for their trustability and commitment. With this SA, we show that a variety of cyberattacks can be mitigated using local trustable resources without stressing the bulk grid. The demonstrations are carried out using a variety of platforms with a high-fidelity co-simulation platform, real-time hardware-in-the-loop validation, and a utility-friendly simulator., Comment: Submitted to the Proceedings of the National Academy of Sciences (PNAS), under review

Evaluating aligned large language models' (LLMs) ability to recognize and reject unsafe user requests is crucial for safe, policy-compliant deployments. Existing evaluation efforts, however, face three limitations that we address with SORRY-Bench, our proposed benchmark. First, existing methods often use coarse-grained taxonomies of unsafe topics, and are over-representing some fine-grained topics. For example, among the ten existing datasets that we evaluated, tests for refusals of self-harm instructions are over 3x less represented than tests for fraudulent activities. SORRY-Bench improves on this by using a fine-grained taxonomy of 45 potentially unsafe topics, and 450 class-balanced unsafe instructions, compiled through human-in-the-loop methods. Second, linguistic characteristics and formatting of prompts are often overlooked, like different languages, dialects, and more -- which are only implicitly considered in many evaluations. We supplement SORRY-Bench with 20 diverse linguistic augmentations to systematically examine these effects. Third, existing evaluations rely on large LLMs (e.g., GPT-4) for evaluation, which can be computationally expensive. We investigate design choices for creating a fast, accurate automated safety evaluator. By collecting 7K+ human annotations and conducting a meta-evaluation of diverse LLM-as-a-judge designs, we show that fine-tuned 7B LLMs can achieve accuracy comparable to GPT-4 scale LLMs, with lower computational cost. Putting these together, we evaluate over 40 proprietary and open-source LLMs on SORRY-Bench, analyzing their distinctive refusal behaviors. We hope our effort provides a building block for systematic evaluations of LLMs' safety refusal capabilities, in a balanced, granular, and efficient manner.

Data Shapley provides a principled framework for attributing data's contribution within machine learning contexts. However, existing approaches require re-training models on different data subsets, which is computationally intensive, foreclosing their application to large-scale models. Furthermore, they produce the same attribution score for any models produced by running the learning algorithm, meaning they cannot perform targeted attribution towards a specific model obtained from a single run of the algorithm. This paper introduces In-Run Data Shapley, which addresses these limitations by offering scalable data attribution for a target model of interest. In its most efficient implementation, our technique incurs negligible additional runtime compared to standard model training. This dramatic efficiency improvement makes it possible to perform data attribution for the foundation model pretraining stage for the first time. We present several case studies that offer fresh insights into pretraining data's contribution and discuss their implications for copyright in generative AI and pretraining data curation.

It is well-known that every transitive linear order is exponentiable. However, is the converse true? This question was posed in Chapter 8 of the textbook titled "Linear Orderings" by Rosenstein. We define the class CTLO of cyclically transitive linear orders that properly contains the class of transitive linear orders, and show that all discrete unbounded orders in CTLO are exponentiable, thereby providing a negative answer to the question. The class CTLO is closely related to the class of transitive cyclic orders introduced by Droste, Giraudet and Macpherson. We also discuss the closure of subclasses of CTLO under products and iterated Hausdorff condensations., Comment: 10 pages

Quantum technology can enable secure communication for cryptography purposes using quantum key distribution. Quantum key distribution protocols provide a secret key between two users with security guaranteed by the laws of quantum mechanics. To define the proper implementation of a quantum key distribution system using a particular cryptography protocol, it is crucial to critically and meticulously assess the device's performance due to technological limitations in the components used. We perform simulations on the ANSYS Interconnect platform to characterise the practical implementation of these devices using distributed-phase-reference protocols differential-phase-shift and coherent-one-way quantum key distribution. Further, we briefly describe and simulate some possible eavesdropping attempts, backflash attack, trojan-horse attack and detector-blinding attack exploiting the device imperfections.

In a recent result, Knop, Lovett, McGuire and Yuan (STOC 2021) proved the log-rank conjecture for communication complexity, up to log n factor, for any Boolean function composed with AND function as the inner gadget. One of the main tools in this result was the relationship between monotone analogues of well-studied Boolean complexity measures like block sensitivity and certificate complexity. The relationship between the standard measures has been a long line of research, with a landmark result by Huang (Annals of Mathematics 2019), finally showing that sensitivity is polynomially related to all other standard measures. In this article, we study the monotone analogues of standard measures like block sensitivity (mbs(f)), certificate complexity (MCC(f)) and fractional block sensitivity (fmbs(f)); and study the relationship between these measures given their connection with AND-decision tree and sparsity of a Boolean function. We show the following results: 1) Given a Boolean function $f : \{0, 1\}^{n} \rightarrow \{0, 1\}$, the ratio $fmbs(f^l )/mbs(f^l )$ is bounded by a function of n (and not l). A similar result was known for the corresponding standard measures (Tal, ITCS 2013). This result allows us to extend any upper bound by a well behaved measure on monotone block sensitivity to monotone fractional block sensitivity. 2) The question of the best possible upper bound on monotone block sensitivity by the logarithm of sparsity is equivalent to the natural question of best upper bound by degree on sensitivity. One side of this relationship was used in the proof by Knop, Lovett, McGuire and Yuan (STOC 2021). 3) For two natural classes of functions, symmetric and monotone, hitting set complexity (MCC) is equal to monotone sensitivity., Comment: 27 Pages, COCOON 2024

The safety alignment of current Large Language Models (LLMs) is vulnerable. Relatively simple attacks, or even benign fine-tuning, can jailbreak aligned models. We argue that many of these vulnerabilities are related to a shared underlying issue: safety alignment can take shortcuts, wherein the alignment adapts a model's generative distribution primarily over only its very first few output tokens. We refer to this issue as shallow safety alignment. In this paper, we present case studies to explain why shallow safety alignment can exist and provide evidence that current aligned LLMs are subject to this issue. We also show how these findings help explain multiple recently discovered vulnerabilities in LLMs, including the susceptibility to adversarial suffix attacks, prefilling attacks, decoding parameter attacks, and fine-tuning attacks. Importantly, we discuss how this consolidated notion of shallow safety alignment sheds light on promising research directions for mitigating these vulnerabilities. For instance, we show that deepening the safety alignment beyond just the first few tokens can often meaningfully improve robustness against some common exploits. Finally, we design a regularized finetuning objective that makes the safety alignment more persistent against fine-tuning attacks by constraining updates on initial tokens. Overall, we advocate that future safety alignment should be made more than just a few tokens deep.

Generative AI has received much attention in the image and language domains, with the transformer neural network continuing to dominate the state of the art. Application of these models to time series generation is less explored, however, and is of great utility to machine learning, privacy preservation, and explainability research. The present survey identifies this gap at the intersection of the transformer, generative AI, and time series data, and reviews works in this sparsely populated subdomain. The reviewed works show great variety in approach, and have not yet converged on a conclusive answer to the problems the domain poses. GANs, diffusion models, state space models, and autoencoders were all encountered alongside or surrounding the transformers which originally motivated the survey. While too open a domain to offer conclusive insights, the works surveyed are quite suggestive, and several recommendations for best practice, and suggestions of valuable future work, are provided.

We examine the sky distribution of radio galaxies in the NRAO VLA Sky Survey (NVSS) and the Rapid ASKAP Continuum Survey (RACS). Analyses of these samples have reported tension between their inferred dipoles and the kinematic dipole of the Cosmic Microwave Background (CMB). This represents a challenge to the traditional assumption that the Universe is homogeneous and isotropic on large scales: the cosmological principle. We find that NVSS and RACS contain local radio sources which give a non-negligible contribution to the overall dipole signal. These need to be adequately accounted for since the aim is to probe the composition of the Universe at large scales. By appropriately considering these sources, the inferred dipole amplitude in either sample is reduced. Nonetheless, we find support for a dipole aligning with that of the CMB but larger in amplitude, especially in the joint analysis. However, the 'clustering dipole' - the contribution of local sources to the net inferred dipole - appears to align with the direction of the CMB dipole, and its magnitude increases as deeper nearby sources are considered up to a comoving distance of $\approx 130$ Mpc ($h=0.7$). The significance of this observation in the context of the cosmological principle is unclear and prompts further inquiry., Comment: 15 pages, 13 figures, accepted for publication in MNRAS

Diffusion models have emerged as effective distribution estimators in vision, language, and reinforcement learning, but their use as priors in downstream tasks poses an intractable posterior inference problem. This paper studies amortized sampling of the posterior over data, $\mathbf{x}\sim p^{\rm post}(\mathbf{x})\propto p(\mathbf{x})r(\mathbf{x})$, in a model that consists of a diffusion generative model prior $p(\mathbf{x})$ and a black-box constraint or likelihood function $r(\mathbf{x})$. We state and prove the asymptotic correctness of a data-free learning objective, relative trajectory balance, for training a diffusion model that samples from this posterior, a problem that existing methods solve only approximately or in restricted cases. Relative trajectory balance arises from the generative flow network perspective on diffusion models, which allows the use of deep reinforcement learning techniques to improve mode coverage. Experiments illustrate the broad potential of unbiased inference of arbitrary posteriors under diffusion priors: in vision (classifier guidance), language (infilling under a discrete diffusion LLM), and multimodal data (text-to-image generation). Beyond generative modeling, we apply relative trajectory balance to the problem of continuous control with a score-based behavior prior, achieving state-of-the-art results on benchmarks in offline reinforcement learning., Comment: Code: https://github.com/GFNOrg/diffusion-finetuning

We introduce PlausiVL, a large video-language model for anticipating action sequences that are plausible in the real-world. While significant efforts have been made towards anticipating future actions, prior approaches do not take into account the aspect of plausibility in an action sequence. To address this limitation, we explore the generative capability of a large video-language model in our work and further, develop the understanding of plausibility in an action sequence by introducing two objective functions, a counterfactual-based plausible action sequence learning loss and a long-horizon action repetition loss. We utilize temporal logical constraints as well as verb-noun action pair logical constraints to create implausible/counterfactual action sequences and use them to train the model with plausible action sequence learning loss. This loss helps the model to differentiate between plausible and not plausible action sequences and also helps the model to learn implicit temporal cues crucial for the task of action anticipation. The long-horizon action repetition loss puts a higher penalty on the actions that are more prone to repetition over a longer temporal window. With this penalization, the model is able to generate diverse, plausible action sequences. We evaluate our approach on two large-scale datasets, Ego4D and EPIC-Kitchens-100, and show improvements on the task of action anticipation., Comment: CVPR 2024

Large autoregressive models like Transformers can solve tasks through in-context learning (ICL) without learning new weights, suggesting avenues for efficiently solving new tasks. For many tasks, e.g., linear regression, the data factorizes: examples are independent given a task latent that generates the data, e.g., linear coefficients. While an optimal predictor leverages this factorization by inferring task latents, it is unclear if Transformers implicitly do so or if they instead exploit heuristics and statistical shortcuts enabled by attention layers. Both scenarios have inspired active ongoing work. In this paper, we systematically investigate the effect of explicitly inferring task latents. We minimally modify the Transformer architecture with a bottleneck designed to prevent shortcuts in favor of more structured solutions, and then compare performance against standard Transformers across various ICL tasks. Contrary to intuition and some recent works, we find little discernible difference between the two; biasing towards task-relevant latent variables does not lead to better out-of-distribution performance, in general. Curiously, we find that while the bottleneck effectively learns to extract latent task variables from context, downstream processing struggles to utilize them for robust prediction. Our study highlights the intrinsic limitations of Transformers in achieving structured ICL solutions that generalize, and shows that while inferring the right latents aids interpretability, it is not sufficient to alleviate this problem.

We give a randomness-efficient homomorphism test in the low soundness regime for functions, $f: G\to \mathbb{U}_t$, from an arbitrary finite group $G$ to $t\times t$ unitary matrices. We show that if such a function passes a derandomized Blum--Luby--Rubinfeld (BLR) test (using small-bias sets), then (i) it correlates with a function arising from a genuine homomorphism, and (ii) it has a non-trivial Fourier mass on a low-dimensional irreducible representation. In the full randomness regime, such a test for matrix-valued functions on finite groups implicitly appears in the works of Gowers and Hatami [Sbornik: Mathematics '17], and Moore and Russell [SIAM Journal on Discrete Mathematics '15]. Thus, our work can be seen as a near-optimal derandomization of their results. Our key technical contribution is a "degree-2 expander mixing lemma'' that shows that Gowers' $\mathrm{U}^2$ norm can be efficiently estimated by restricting it to a small-bias subset. Another corollary is a "derandomized'' version of a useful lemma due to Babai, Nikolov, and Pyber [SODA'08].

The exposure of security vulnerabilities in safety-aligned language models, e.g., susceptibility to adversarial attacks, has shed light on the intricate interplay between AI safety and AI security. Although the two disciplines now come together under the overarching goal of AI risk management, they have historically evolved separately, giving rise to differing perspectives. Therefore, in this paper, we advocate that stakeholders in AI risk management should be aware of the nuances, synergies, and interplay between safety and security, and unambiguously take into account the perspectives of both disciplines in order to devise mostly effective and holistic risk mitigation approaches. Unfortunately, this vision is often obfuscated, as the definitions of the basic concepts of "safety" and "security" themselves are often inconsistent and lack consensus across communities. With AI risk management being increasingly cross-disciplinary, this issue is particularly salient. In light of this conceptual challenge, we introduce a unified reference framework to clarify the differences and interplay between AI safety and AI security, aiming to facilitate a shared understanding and effective collaboration across communities.

How can Large Language Models (LLMs) be aligned with human intentions and values? A typical solution is to gather human preference on model outputs and finetune the LLMs accordingly while ensuring that updates do not deviate too far from a reference model. Recent approaches, such as direct preference optimization (DPO), have eliminated the need for unstable and sluggish reinforcement learning optimization by introducing close-formed supervised losses. However, a significant limitation of the current approach is its design for a single reference model only, neglecting to leverage the collective power of numerous pretrained LLMs. To overcome this limitation, we introduce a novel closed-form formulation for direct preference optimization using multiple reference models. The resulting algorithm, Multi-Reference Preference Optimization (MRPO), leverages broader prior knowledge from diverse reference models, substantially enhancing preference learning capabilities compared to the single-reference DPO. Our experiments demonstrate that LLMs finetuned with MRPO generalize better in various preference data, regardless of data scarcity or abundance. Furthermore, MRPO effectively finetunes LLMs to exhibit superior performance in several downstream natural language processing tasks such as GSM8K and TruthfulQA., Comment: 20 pages

Retrieval-augmented generation (RAG) has been shown vulnerable to retrieval corruption attacks: an attacker can inject malicious passages into retrieval results to induce inaccurate responses. In this paper, we propose RobustRAG as the first defense framework against retrieval corruption attacks. The key insight of RobustRAG is an isolate-then-aggregate strategy: we get LLM responses from each passage in isolation and then securely aggregate these isolated responses. To instantiate RobustRAG, we design keyword-based and decoding-based algorithms for securely aggregating unstructured text responses. Notably, RobustRAG can achieve certifiable robustness: we can formally prove and certify that, for certain queries, RobustRAG can always return accurate responses, even when the attacker has full knowledge of our defense and can arbitrarily inject a small number of malicious passages. We evaluate RobustRAG on open-domain QA and long-form text generation datasets and demonstrate its effectiveness and generalizability across various tasks and datasets.

Deep-sea archives that include intermediate-lived radioactive $^{60}\mathrm{Fe}$ particles suggest the occurrence of several recent supernovae inside the present-day volume of the Local Bubble during the last $\sim 10$ Myr. The isotope $^{60}\mathrm{Fe}$ is mainly produced in massive stars and ejected in supernova explosions, which should always result in a sizeable yield of $^{26}\mathrm{Al}$ from the same objects. $^{60}\mathrm{Fe}$ and $^{26}\mathrm{Al}$ decay with lifetimes of 3.82 and 1.05 Myr, and emit $\gamma$-rays at 1332 and 1809 keV, respectively. These $\gamma$-rays have been measured as diffuse glow of the Milky Way, and would also be expected from inside the Local Bubble as foreground emission. Based on two scenarios, one employing a geometrical model and the other state-of-the-art hydrodynamics simulations, we estimate the expected fluxes of the 1332 and 1809 keV $\gamma$-ray lines, as well as the resulting 511 keV line from positron annihilation due to the $^{26}\mathrm{Al}$ $\beta^+$-decay. We find fluxes in the range of $10^{-6}$-$10^{-5}\,\mathrm{ph\,cm^{-2}\,s^{-1}}$ for all three lines with isotropic contributions of 10-50%. We show that these fluxes are within reach for the upcoming COSI-SMEX $\gamma$-ray telescope over its nominal satellite mission duration of 2 yr. Given the Local Bubble models considered, we conclude that in the case of 10-20 Myr-old superbubbles, the distributions of $^{60}\mathrm{Fe}$ and $^{26}\mathrm{Al}$ are not co-spatial - an assumption usually made in $\gamma$-ray data analyses. In fact, this should be taken into account however when analysing individual nearby targets for their $^{60}\mathrm{Fe}$ to $^{26}\mathrm{Al}$ flux ratio as this gauges the stellar evolution models and the age of the superbubbles. A flux ratio measured for the Local Bubble could further constrain models of $^{60}\mathrm{Fe}$ deposition on Earth and its moon., Comment: accepted in A&A, 18 pages, 13 figures, 2 tables

The following question arises naturally in the study of graph streaming algorithms: "Is there any graph problem which is "not too hard", in that it can be solved efficiently with total communication (nearly) linear in the number $n$ of vertices, and for which, nonetheless, any streaming algorithm with $\tilde{O}(n)$ space (i.e., a semi-streaming algorithm) needs a polynomial $n^{\Omega(1)}$ number of passes?" Assadi, Chen, and Khanna [STOC 2019] were the first to prove that this is indeed the case. However, the lower bounds that they obtained are for rather non-standard graph problems. Our first main contribution is to present the first polynomial-pass lower bounds for natural "not too hard" graph problems studied previously in the streaming model: $k$-cores and degeneracy. We devise a novel communication protocol for both problems with near-linear communication, thus showing that $k$-cores and degeneracy are natural examples of "not too hard" problems. Indeed, previous work have developed single-pass semi-streaming algorithms for approximating these problems. In contrast, we prove that any semi-streaming algorithm for exactly solving these problems requires (almost) $\Omega(n^{1/3})$ passes. Our second main contribution is improved round-communication lower bounds for the underlying communication problems at the basis of these reductions: * We improve the previous lower bound of Assadi, Chen, and Khanna for hidden pointer chasing (HPC) to achieve optimal bounds. * We observe that all current reductions from HPC can also work with a generalized version of this problem that we call MultiHPC, and prove an even stronger and optimal lower bound for this generalization. These two results collectively allow us to improve the resulting pass lower bounds for semi-streaming algorithms by a polynomial factor, namely, from $n^{1/5}$ to $n^{1/3}$ passes., Comment: Accepted at CCC 2024

The low carbon content of Earth and primitive meteorites compared to the Sun and interstellar grains suggests that carbon-rich grains were destroyed in the inner few astronomical units of the young solar system. A promising mechanism to selectively destroy carbonaceous grains is thermal sublimation within the soot line at $\gtrsim$ 300 K. To address whether such hot conditions are common amongst low-mass protostars, we observe CH$_3$CN transitions at 1, 2 and 3 mm with the NOrthern Extended Millimeter Array (NOEMA) toward seven low-mass and one intermediate-mass protostar ($L_{\rm{bol}} \sim2-300 L_\odot$), as CH$_3$CN is an excellent temperature tracer. We find $>$ 300 K gas toward all sources, indicating that hot gas may be prevalent. Moreover, the excitation temperature for CH$_3$OH obtained with the same observations is always lower ($\sim$135-250 K), suggesting that CH$_3$CN and CH$_3$OH have a different spatial distribution. A comparison of the column densities at 1 and 3 mm shows a stronger increase at 3 mm for CH$_3$CN than for CH$_3$OH. Since the dust opacity is lower at longer wavelengths, this indicates that CH$_3$CN is enhanced in the hot gas compared to CH$_3$OH. If this CH$_3$CN enhancement is the result of carbon-grain sublimation, these results suggests that Earth's initial formation conditions may not be rare., Comment: 19 pages, 9 figures, 4 tables (plus 13 pages appendix with 10 figures, 4 tables). Accepted for publication in ApJ

This paper presents BARKPLUG V.2, a Large Language Model (LLM)-based chatbot system built using Retrieval Augmented Generation (RAG) pipelines to enhance the user experience and access to information within academic settings.The objective of BARKPLUG V.2 is to provide information to users about various campus resources, including academic departments, programs, campus facilities, and student resources at a university setting in an interactive fashion. Our system leverages university data as an external data corpus and ingests it into our RAG pipelines for domain-specific question-answering tasks. We evaluate the effectiveness of our system in generating accurate and pertinent responses for Mississippi State University, as a case study, using quantitative measures, employing frameworks such as Retrieval Augmented Generation Assessment(RAGAS). Furthermore, we evaluate the usability of this system via subjective satisfaction surveys using the System Usability Scale (SUS). Our system demonstrates impressive quantitative performance, with a mean RAGAS score of 0.96, and experience, as validated by usability assessments.

Modern cloud has turned data services into easily accessible commodities. With just a few clicks, users are now able to access a catalog of data processing systems for a wide range of tasks. However, the cloud brings in both complexity and opportunity. While cloud users can quickly start an application by using various data services, it can be difficult to configure and optimize these services to gain the most value from them. For cloud providers, managing every aspect of an ever-increasing set of data services, while meeting customer SLAs and minimizing operational cost is becoming more challenging. Cloud technology enables the collection of significant amounts of workload traces and system telemetry. With the progress in data science (DS) and machine learning (ML), it is feasible and desirable to utilize a data-driven, ML-based approach to automate various aspects of data services, resulting in the creation of autonomous data services. This paper presents our perspectives and insights on creating autonomous data services on Azure. It also covers the future endeavors we plan to undertake and unresolved issues that still need attention., Comment: SIGMOD Companion of the 2023 International Conference on Management of Data. 2023

Current research on defending against adversarial examples focuses primarily on achieving robustness against a single attack type such as $\ell_2$ or $\ell_{\infty}$-bounded attacks. However, the space of possible perturbations is much larger and currently cannot be modeled by a single attack type. The discrepancy between the focus of current defenses and the space of attacks of interest calls to question the practicality of existing defenses and the reliability of their evaluation. In this position paper, we argue that the research community should look beyond single attack robustness, and we draw attention to three potential directions involving robustness against multiple attacks: simultaneous multiattack robustness, unforeseen attack robustness, and a newly defined problem setting which we call continual adaptive robustness. We provide a unified framework which rigorously defines these problem settings, synthesize existing research in these fields, and outline open directions. We hope that our position paper inspires more research in simultaneous multiattack, unforeseen attack, and continual adaptive robustness.

We study the communication complexity of $(\Delta + 1)$ vertex coloring, where the edges of an $n$-vertex graph of maximum degree $\Delta$ are partitioned between two players. We provide a randomized protocol which uses $O(n)$ bits of communication and ends with both players knowing the coloring. Combining this with a folklore $\Omega(n)$ lower bound, this settles the randomized communication complexity of $(\Delta + 1)$-coloring up to constant factors., Comment: 16 pages, 1 figure; full version of paper accepted to PODC '24

For green AI, it is crucial to measure and reduce the carbon footprint emitted during the training of large language models. In NLP, performing pre-training on Transformer models requires significant computational resources. This pre-training involves using a large amount of text data to gain prior knowledge for performing downstream tasks. Thus, it is important that we select the correct data in the form of domain-specific data from this vast corpus to achieve optimum results aligned with our domain-specific tasks. While training on large unsupervised data is expensive, it can be optimized by performing a data selection step before pretraining. Selecting important data reduces the space overhead and the substantial amount of time required to pre-train the model while maintaining constant accuracy. We investigate the existing selection strategies and propose our own domain-adaptive data selection method - TextGram - that effectively selects essential data from large corpora. We compare and evaluate the results of finetuned models for text classification task with and without data selection. We show that the proposed strategy works better compared to other selection methods., Comment: Accepted at SPELLL 2023

The availability of text or topic classification datasets in the low-resource Marathi language is limited, typically consisting of fewer than 4 target labels, with some achieving nearly perfect accuracy. In this work, we introduce L3Cube-MahaNews, a Marathi text classification corpus that focuses on News headlines and articles. This corpus stands out as the largest supervised Marathi Corpus, containing over 1.05L records classified into a diverse range of 12 categories. To accommodate different document lengths, MahaNews comprises three supervised datasets specifically designed for short text, long documents, and medium paragraphs. The consistent labeling across these datasets facilitates document length-based analysis. We provide detailed data statistics and baseline results on these datasets using state-of-the-art pre-trained BERT models. We conduct a comparative analysis between monolingual and multilingual BERT models, including MahaBERT, IndicBERT, and MuRIL. The monolingual MahaBERT model outperforms all others on every dataset. These resources also serve as Marathi topic classification datasets or models and are publicly available at https://github.com/l3cube-pune/MarathiNLP ., Comment: Accepted at SPELLL 2023

Physics-based simulations have accelerated progress in robot learning for driving, manipulation, and locomotion. Yet, a fast, accurate, and robust surgical simulation environment remains a challenge. In this paper, we present ORBIT-Surgical, a physics-based surgical robot simulation framework with photorealistic rendering in NVIDIA Omniverse. We provide 14 benchmark surgical tasks for the da Vinci Research Kit (dVRK) and Smart Tissue Autonomous Robot (STAR) which represent common subtasks in surgical training. ORBIT-Surgical leverages GPU parallelization to train reinforcement learning and imitation learning algorithms to facilitate study of robot learning to augment human surgical skills. ORBIT-Surgical also facilitates realistic synthetic data generation for active perception tasks. We demonstrate ORBIT-Surgical sim-to-real transfer of learned policies onto a physical dVRK robot. Project website: orbit-surgical.github.io

We conduct a preliminary study of the effect of GPT's temperature parameter on the diversity of GPT4-generated questions. We find that using higher temperature values leads to significantly higher diversity, with different temperatures exposing different types of similarity between generated sets of questions. We also demonstrate that diverse question generation is especially difficult for questions targeting lower levels of Bloom's Taxonomy.

Data augmentation is an important facilitator of deep learning applications in the time series domain. A gap is identified in the literature, demonstrating sparse exploration of the transformer, the dominant sequence model, for data augmentation in time series. A architecture hybridizing several successful priors is put forth and tested using a powerful time domain similarity metric. Results suggest the challenge of this domain, and several valuable directions for future work.

Locational measures of accessibility are widely used in urban and transportation planning to understand the impact of the transportation system on influencing people's access to places. However, there is a considerable lack of measurement standards and publicly available data. We propose a generalized measure of locational accessibility that has a comprehensible form for transportation planning analysis. This metric combines the cumulative opportunities approach with gravity-based measures and is capable of catering to multiple trip purposes, travel modes, cost thresholds, and scales of analysis. Using data from multiple publicly available datasets, this metric is computed by trip purpose and travel time threshold for all block groups in the United States, and the data is made publicly accessible. Further, case studies of three large metropolitan areas reveal substantial inefficiencies in transportation infrastructure, with the most inefficiency observed in sprawling and non-core urban areas, especially for bicycling. Subsequently, it is shown that targeted investment in facilities can contribute to a more equitable distribution of accessibility to essential shopping and service facilities. By assigning greater weights to socioeconomically disadvantaged neighborhoods, the proposed metric formally incorporates equity considerations into transportation planning, contributing to a more equitable distribution of accessibility to essential services and facilities., Comment: Article originally submitted to the Journal of Transport Geography. Article contains 5 main figures. Supplementary Material is prepared but not uploaded in this submission