Search

Personalization of Large Language Models (LLMs) has recently become increasingly important with a wide range of applications. Despite the importance and recent progress, most existing works on personalized LLMs have focused either entirely on (a) personalized text generation or (b) leveraging LLMs for personalization-related downstream applications, such as recommendation systems. In this work, we bridge the gap between these two separate main directions for the first time by introducing a taxonomy for personalized LLM usage and summarizing the key differences and challenges. We provide a formalization of the foundations of personalized LLMs that consolidates and expands notions of personalization of LLMs, defining and discussing novel facets of personalization, usage, and desiderata of personalized LLMs. We then unify the literature across these diverse fields and usage scenarios by proposing systematic taxonomies for the granularity of personalization, personalization techniques, datasets, evaluation methods, and applications of personalized LLMs. Finally, we highlight challenges and important open problems that remain to be addressed. By unifying and surveying recent research using the proposed taxonomies, we aim to provide a clear guide to the existing literature and different facets of personalization in LLMs, empowering both researchers and practitioners.

Blazars are beamed active galactic nuclei known for their strong multi-wavelength variability on timescales from years down to minutes. We aim to investigate the suitability of the twisting jet model presented in previous works to explain the multi-wavelength behaviour of BL Lacertae, the prototype of one of the blazar classes. According to this model, the jet is inhomogeneous, curved, and twisting, and the long-term variability is due to changes in the Doppler factor due to variations in the orientation of the jet-emitting regions. We analysed optical data of the source obtained during monitoring campaigns organised by the Whole Earth Blazar Telescope (WEBT) in 2019-2022, together with radio data from the WEBT and other teams, and gamma-ray data from the Fermi satellite. In this period, BL Lacertae underwent an extraordinary activity phase, reaching its historical optical and gamma-ray brightness maxima. The application of the twisting jet model to the source light curves allows us to infer the wiggling motion of the optical, radio, and gamma-ray jet-emitting regions. The optical-radio correlation shows that the changes in the radio viewing angle follow those in the optical viewing angle by about 120 days, and it suggests that the jet is composed of plasma filaments, which is in agreement with some radio high-resolution observations of other sources. The gamma-ray emitting region is found to be co-spatial with the optical one, and the analysis of the gamma-optical correlation is consistent with both the geometric interpretation and a synchrotron self-Compton (SSC) origin of the high-energy photons. We propose a geometric scenario where the jet is made up of a pair of emitting plasma filaments in a sort of double-helix curved rotating structure, whose wiggling motion produces changes in the Doppler beaming and can thus explain the observed multi-wavelength long-term variability., Comment: In press for A&A

The reproducibility of scientific articles is central to the advancement of science. Despite this importance, evaluating reproducibility remains challenging due to the scarcity of ground truth data. Predictive models can address this limitation by streamlining the tedious evaluation process. Typically, a paper's reproducibility is inferred based on the availability of artifacts such as code, data, or supplemental information, often without extensive empirical investigation. To address these issues, we utilized artifacts of papers as fundamental units to develop a novel, dual-spectrum framework that focuses on author-centric and external-agent perspectives. We used the author-centric spectrum, followed by the external-agent spectrum, to guide a structured, model-based approach to quantify and assess reproducibility. We explored the interdependencies between different factors influencing reproducibility and found that linguistic features such as readability and lexical diversity are strongly correlated with papers achieving the highest statuses on both spectrums. Our work provides a model-driven pathway for evaluating the reproducibility of scientific research. The code, methods, and artifacts for our study are publicly available at: https://github.com/reproducibilityproject/NLRR/, Comment: 10 pages, CIKM 2024

Here we present several use cases for using Generative AI (Gen AI) to improve systems engineering and cognitive knowledge management related to the future of astronomy from a culmination of working meetings and presentations as part of the Gen AI Task Group for the NASA Habitable Worlds Observatory (HWO) Science and Technology Architecture Review Team (START) AI/ML Working Group. Collectively, our group mission statement is "Where is the Human-in-the-loop as Gen AI systems become more powerful and autonomous?" with an emphasis on the ethical applications of Gen AI, guided by using these systems to remove drudgery from human work while simultaneously increasing opportunities for humans to experience more collective creativity and innovation. The HWO mission stands to benefit dramatically from generative models for different data types including text, time series/spectra, and image data. These cover a wide range of applications in science and engineering for HWO, including: mission development acceleration, data analysis and interpretation, enhancing imaging capabilities, anomaly detection, predictive modeling and simulation, data augmentation for machine learning, instrument calibration and optimization, public engagement and education, and assisting in mission planning. As an example, through sensitivity analysis of simulated exoplanet population science data sets of various generative model complexity, we can reverse engineer the measurement uncertainty requirements for HWO instruments to produce data that can constrain population models and thus inform HWO design requirements. This approach to HWO design is one example of a strategy that can ensure that HWO remains AI-ready. Through presenting herein a combination of visionary ideas balanced with grounded validated use case examples, we aim to support the development of a long-term strategy to keep HWO AI-ready as it moves forward., Comment: Lack of guidelines for submitting work that came out of the HWO START TAG working groups.

State Estimation is a crucial task in power systems. Graph Neural Networks have demonstrated significant potential in state estimation for power systems by effectively analyzing measurement data and capturing the complex interactions and interrelations among the measurements through the system's graph structure. However, the information about the system's graph structure may be inaccurate due to noise, attack or lack of accurate information about the topology of the system. This paper studies these scenarios under topology uncertainties and evaluates the impact of the topology uncertainties on the performance of a Temporal Graph Convolutional Network (TGCN) for state estimation in power systems. In order to make the model resilient to topology uncertainties, modifications in the TGCN model are proposed to incorporate a knowledge graph, generated based on the measurement data. This knowledge graph supports the assumed uncertain system graph. Two variations of the TGCN architecture are introduced to integrate the knowledge graph, and their performances are evaluated and compared to demonstrate improved resilience against topology uncertainties. The evaluation results indicate that while the two proposed architecture show different performance, they both improve the performance of the TGCN state estimation under topology uncertainties., Comment: 9 pages, 5 figures

This paper proposes a novel distributed approach for solving a cooperative Constrained Multi-agent Reinforcement Learning (CMARL) problem, where agents seek to minimize a global objective function subject to shared constraints. Unlike existing methods that rely on centralized training or coordination, our approach enables fully decentralized online learning, with each agent maintaining local estimates of both primal and dual variables. Specifically, we develop a distributed primal-dual algorithm based on actor-critic methods, leveraging local information to estimate Lagrangian multipliers. We establish consensus among the Lagrangian multipliers across agents and prove the convergence of our algorithm to an equilibrium point, analyzing the sub-optimality of this equilibrium compared to the exact solution of the unparameterized problem. Furthermore, we introduce a constrained cooperative Cournot game with stochastic dynamics as a test environment to evaluate the algorithm's performance in complex, real-world scenarios.

Large language models (LLMs) have shown impressive potential in helping with numerous medical challenges. Deploying LLMs in high-stakes applications such as medicine, however, brings in many concerns. One major area of concern relates to biased behaviors of LLMs in medical applications, leading to unfair treatment of individuals. To pave the way for the responsible and impactful deployment of Med LLMs, rigorous evaluation is a key prerequisite. Due to the huge complexity and variability of different medical scenarios, existing work in this domain has primarily relied on using manually crafted datasets for bias evaluation. In this study, we present a new method to scale up such bias evaluations by automatically generating test cases based on rigorous medical evidence. We specifically target the challenges of a) domain-specificity of bias characterization, b) hallucinating while generating the test cases, and c) various dependencies between the health outcomes and sensitive attributes. To that end, we offer new methods to address these challenges integrated with our generative pipeline, using medical knowledge graphs, medical ontologies, and customized general LLM evaluation frameworks in our method. Through a series of extensive experiments, we show that the test cases generated by our proposed method can effectively reveal bias patterns in Med LLMs at larger and more flexible scales than human-crafted datasets. We publish a large bias evaluation dataset using our pipeline, which is dedicated to a few medical case studies. A live demo of our application for vignette generation is available at https://vignette.streamlit.app. Our code is also available at https://github.com/healthylaife/autofair.

The recent introduction of large language models (LLMs) has revolutionized the field of robotics by enabling contextual reasoning and intuitive human-robot interaction in domains as varied as manipulation, locomotion, and self-driving vehicles. When viewed as a stand-alone technology, LLMs are known to be vulnerable to jailbreaking attacks, wherein malicious prompters elicit harmful text by bypassing LLM safety guardrails. To assess the risks of deploying LLMs in robotics, in this paper, we introduce RoboPAIR, the first algorithm designed to jailbreak LLM-controlled robots. Unlike existing, textual attacks on LLM chatbots, RoboPAIR elicits harmful physical actions from LLM-controlled robots, a phenomenon we experimentally demonstrate in three scenarios: (i) a white-box setting, wherein the attacker has full access to the NVIDIA Dolphins self-driving LLM, (ii) a gray-box setting, wherein the attacker has partial access to a Clearpath Robotics Jackal UGV robot equipped with a GPT-4o planner, and (iii) a black-box setting, wherein the attacker has only query access to the GPT-3.5-integrated Unitree Robotics Go2 robot dog. In each scenario and across three new datasets of harmful robotic actions, we demonstrate that RoboPAIR, as well as several static baselines, finds jailbreaks quickly and effectively, often achieving 100% attack success rates. Our results reveal, for the first time, that the risks of jailbroken LLMs extend far beyond text generation, given the distinct possibility that jailbroken robots could cause physical damage in the real world. Indeed, our results on the Unitree Go2 represent the first successful jailbreak of a deployed commercial robotic system. Addressing this emerging vulnerability is critical for ensuring the safe deployment of LLMs in robotics. Additional media is available at: https://robopair.org

The study presents a novel approach for stochastic nonlinear model updating in structural dynamics, employing a Bayesian framework integrated with Markov Chain Monte Carlo (MCMC) sampling for parameter estimation by using an approximated likelihood function. The proposed methodology is applied to both numerical and experimental cases. The paper commences by introducing Bayesian inference and its constituents: the likelihood function, prior distribution, and posterior distribution. The resonant decay method is employed to extract backbone curves, which capture the non-linear behaviour of the system. A mathematical model based on a single degree of freedom (SDOF) system is formulated, and backbone curves are obtained from time response data. Subsequently, MCMC sampling is employed to estimate the parameters using both numerical and experimental data. The obtained results demonstrate the convergence of the Markov chain, present parameter trace plots, and provide estimates of posterior distributions of updated parameters along with their uncertainties. Experimental validation is performed on a cantilever beam system equipped with permanent magnets and electromagnets. The proposed methodology demonstrates promising results in estimating parameters of stochastic non-linear dynamical systems. Through the use of the proposed likelihood functions using backbone curves, the probability distributions of both linear and non-linear parameters are simultaneously identified. Based on this view, the necessity to segregate stochastic linear and non-linear model updating is eliminated.

This paper investigates the design of a unified search engine to serve multiple retrieval-augmented generation (RAG) agents, each with a distinct task, backbone large language model (LLM), and retrieval-augmentation strategy. We introduce an iterative approach where the search engine generates retrieval results for these RAG agents and gathers feedback on the quality of the retrieved documents during an offline phase. This feedback is then used to iteratively optimize the search engine using a novel expectation-maximization algorithm, with the goal of maximizing each agent's utility function. Additionally, we adapt this approach to an online setting, allowing the search engine to refine its behavior based on real-time individual agents feedback to better serve the results for each of them. Experiments on diverse datasets from the Knowledge-Intensive Language Tasks (KILT) benchmark demonstrates that our approach significantly on average outperforms competitive baselines across 18 RAG models. We also demonstrate that our method effectively ``personalizes'' the retrieval process for each RAG agent based on the collected feedback. Finally, we provide a comprehensive ablation study to explore various aspects of our method.

The goal of this paper is to improve (upcycle) an existing large language model without the prohibitive requirements of continued pre-training of the full-model. The idea is to split the pre-training data into semantically relevant groups and train an expert on each subset. An expert takes the form of a lightweight adapter added on the top of a frozen base model. During inference, an incoming query is first routed to the most relevant expert which is then loaded onto the base model for the forward pass. Unlike typical Mixture of Experts (MoE) models, the experts in our method do not work with other experts for a single query. Hence, we dub them "introvert" experts. Freezing the base model and keeping the experts as lightweight adapters allows extreme parallelism during training and inference. Training of all experts can be done in parallel without any communication channels between them. Similarly, the inference can also be heavily parallelized by distributing experts on different GPUs and routing each request to the GPU containing its relevant expert. We implement a proof-of-concept version of this method and show the validity of our approach.

Thin-film lithium niobate (TFLN) on insulator is a promising platform for nonlinear photonic integrated circuits (PICs) due to its strong light confinement, high second-order nonlinearity, and flexible quasi-phase matching for three-wave mixing processes via periodic poling. Among the three-wave mixing processes of interest, difference frequency generation (DFG) can produce long wave infrared (IR) light from readily available near IR inputs. While broadband DFG is well studied for mid-IR frequencies, achieving broadband idler generation within the telecom window (near C-band) and the short-wave infrared (near 2 micron) is more challenging due to stringent dispersion profile requirements, especially when using standardized TFLN thicknesses. In this paper, we investigate various standard waveguide designs to pinpoint favorable conditions for broadband DFG operation covering several telecom bands. Our simulations identify viable designs with a possible 3-dB conversion efficiency bandwidth (CE-BW) of 300 nm and our measurements show idler generation from 1418 nm to 1740 nm, limited by our available sources, experimentally confirming our design approach. Furthermore, temperature tuning allows further shift of the idler towards the mid-IR, up to 1819 nm. We also achieve a stretched wavelength range of idler generation by leveraging the longitudinal variation of the waveguide in addition to poling. Finally, our numerical simulations show the possibility of extending the CE-BW up to 780 nm while focusing on waveguide cross-sections that are available for fabrication within a foundry. Our work provides a methodology that bridges the deviations between fabricated and designed cross-sections, paving a way for standardized broadband DFG building blocks.

6G, the next generation of mobile networks, is set to offer even higher data rates, ultra-reliability, and lower latency than 5G. New 6G services will increase the load and dynamism of the network. Network Function Virtualization (NFV) aids with this increased load and dynamism by eliminating hardware dependency. It aims to boost the flexibility and scalability of network deployment services by separating network functions from their specific proprietary forms so that they can run as virtual network functions (VNFs) on commodity hardware. It is essential to design an NFV orchestration and management framework to support these services. However, deploying bulky monolithic VNFs on the network is difficult, especially when underlying resources are scarce, resulting in ineffective resource management. To address this, microservices-based NFV approaches are proposed. In this approach, monolithic VNFs are decomposed into micro VNFs, increasing the likelihood of their successful placement and resulting in more efficient resource management. This article discusses the proposed framework for resource allocation for microservices-based services to provide end-to-end Quality of Service (QoS) using the Double Deep Q Learning (DDQL) approach. Furthermore, to enhance this resource allocation approach, we discussed and addressed two crucial sub-problems: the need for a dynamic priority technique and the presence of the low-priority starvation problem. Using the Deep Deterministic Policy Gradient (DDPG) model, an Adaptive Scheduling model is developed that effectively mitigates the starvation problem. Additionally, the impact of incorporating traffic load considerations into deployment and scheduling is thoroughly investigated.

State-of-the-art hypergraph partitioners utilize a multilevel paradigm to construct progressively coarser hypergraphs across multiple layers, guiding cut refinements at each level of the hierarchy. Traditionally, these partitioners employ heuristic methods for coarsening and do not consider the structural features of hypergraphs. In this work, we introduce a multilevel spectral framework, SHyPar, for partitioning large-scale hypergraphs by leveraging hyperedge effective resistances and flow-based community detection techniques. Inspired by the latest theoretical spectral clustering frameworks, such as HyperEF and HyperSF, SHyPar aims to decompose large hypergraphs into multiple subgraphs with few inter-partition hyperedges (cut size). A key component of SHyPar is a flow-based local clustering scheme for hypergraph coarsening, which incorporates a max-flow-based algorithm to produce clusters with substantially improved conductance. Additionally, SHyPar utilizes an effective resistance-based rating function for merging nodes that are strongly connected (coupled). Compared with existing state-of-the-art hypergraph partitioning methods, our extensive experimental results on real-world VLSI designs demonstrate that SHyPar can more effectively partition hypergraphs, achieving state-of-the-art solution quality., Comment: 14 pages, 11 figures, 4 tables

Evaluating the creativity of large language models (LLMs) in story writing is difficult because LLM-generated stories could seemingly look creative but be very similar to some existing stories in their huge and proprietary training corpus. To overcome this challenge, we introduce a novel benchmark dataset with varying levels of prompt specificity: CS4 ($\mathbf{C}$omparing the $\mathbf{S}$kill of $\mathbf{C}$reating $\mathbf{S}$tories by $\mathbf{C}$ontrolling the $\mathbf{S}$ynthesized $\mathbf{C}$onstraint $\mathbf{S}$pecificity). By increasing the number of requirements/constraints in the prompt, we can increase the prompt specificity and hinder LLMs from retelling high-quality narratives in their training data. Consequently, CS4 empowers us to indirectly measure the LLMs' creativity without human annotations. Our experiments on LLaMA, Gemma, and Mistral not only highlight the creativity challenges LLMs face when dealing with highly specific prompts but also reveal that different LLMs perform very differently under different numbers of constraints and achieve different balances between the model's instruction-following ability and narrative coherence. Additionally, our experiments on OLMo suggest that Learning from Human Feedback (LHF) can help LLMs select better stories from their training data but has limited influence in boosting LLMs' ability to produce creative stories that are unseen in the training corpora. The benchmark is released at https://github.com/anirudhlakkaraju/cs4_benchmark.

Machine unlearning (MU) aims to remove the influence of particular data points from the learnable parameters of a trained machine learning model. This is a crucial capability in light of data privacy requirements, trustworthiness, and safety in deployed models. MU is particularly challenging for deep neural networks (DNNs), such as convolutional nets or vision transformers, as such DNNs tend to memorize a notable portion of their training dataset. Nevertheless, the community lacks a rigorous and multifaceted study that looks into the success of MU methods for DNNs. In this paper, we investigate 18 state-of-the-art MU methods across various benchmark datasets and models, with each evaluation conducted over 10 different initializations, a comprehensive evaluation involving MU over 100K models. We show that, with the proper hyperparameters, Masked Small Gradients (MSG) and Convolution Transpose (CT), consistently perform better in terms of model accuracy and run-time efficiency across different models, datasets, and initializations, assessed by population-based membership inference attacks (MIA) and per-sample unlearning likelihood ratio attacks (U-LiRA). Furthermore, our benchmark highlights the fact that comparing a MU method only with commonly used baselines, such as Gradient Ascent (GA) or Successive Random Relabeling (SRL), is inadequate, and we need better baselines like Negative Gradient Plus (NG+) with proper hyperparameter selection.

Despite significant advancements, Large Language Models (LLMs) exhibit blind spots that impair their ability to retrieve and process relevant contextual data effectively. We demonstrate that LLM performance in graph tasks with complexities beyond the "needle-in-a-haystack" scenario-where solving the problem requires cross-referencing and reasoning across multiple subproblems jointly-is influenced by the proximity of relevant information within the context, a phenomenon we term "lost-in-distance". We examine two fundamental graph tasks: identifying common connections between two nodes and assessing similarity among three nodes, and show that the model's performance in these tasks significantly depends on the relative positioning of common edges. We evaluate three publicly available LLMs-Llama-3-8B, Llama-3-70B, and GPT-4-using various graph encoding techniques that represent graph structures for LLM input. We propose a formulation for the lost-in-distance phenomenon and demonstrate that lost-in-distance and lost-in-the middle phenomenas occur independently. Results indicate that model accuracy can decline by up to 6x as the distance between node connections increases, independent of graph encoding and model size.

Model Interpretation aims at the extraction of insights from the internals of a trained model. A common approach to address this task is the characterization of relevant features internally encoded in the model that are critical for its proper operation. Despite recent progress of these methods, they come with the weakness of being computationally expensive due to the dense evaluation of datasets that they require. As a consequence, research on the design of these methods have focused on smaller data subsets which may led to reduced insights. To address these computational costs, we propose a coreset-based interpretation framework that utilizes coreset selection methods to sample a representative subset of the large dataset for the interpretation task. Towards this goal, we propose a similarity-based evaluation protocol to assess the robustness of model interpretation methods towards the amount data they take as input. Experiments considering several interpretation methods, DNN models, and coreset selection methods show the effectiveness of the proposed framework.

With the remarkable capability to reach the public instantly, social media has become integral in sharing scholarly articles to measure public response. Since spamming by bots on social media can steer the conversation and present a false public interest in given research, affecting policies impacting the public's lives in the real world, this topic warrants critical study and attention. We used the Altmetric dataset in combination with data collected through the Twitter Application Programming Interface (API) and the Botometer API. We combined the data into an extensive dataset with academic articles, several features from the article and a label indicating whether the article had excessive bot activity on Twitter or not. We analyzed the data to see the possibility of bot activity based on different characteristics of the article. We also trained machine-learning models using this dataset to identify possible bot activity in any given article. Our machine-learning models were capable of identifying possible bot activity in any academic article with an accuracy of 0.70. We also found that articles related to "Health and Human Science" are more prone to bot activity compared to other research areas. Without arguing the maliciousness of the bot activity, our work presents a tool to identify the presence of bot activity in the dissemination of an academic article and creates a baseline for future research in this direction., Comment: 38 pages, 10 figures. Aslib Journal of Information Management, Vol. ahead-of-print No. ahead-of-print

In response to the growing complexity and volume of scientific literature, this paper introduces the LLMs4Synthesis framework, designed to enhance the capabilities of Large Language Models (LLMs) in generating high-quality scientific syntheses. This framework addresses the need for rapid, coherent, and contextually rich integration of scientific insights, leveraging both open-source and proprietary LLMs. It also examines the effectiveness of LLMs in evaluating the integrity and reliability of these syntheses, alleviating inadequacies in current quantitative metrics. Our study contributes to this field by developing a novel methodology for processing scientific papers, defining new synthesis types, and establishing nine detailed quality criteria for evaluating syntheses. The integration of LLMs with reinforcement learning and AI feedback is proposed to optimize synthesis quality, ensuring alignment with established criteria. The LLMs4Synthesis framework and its components are made available, promising to enhance both the generation and evaluation processes in scientific research synthesis., Comment: 12 pages, 3 figures, Accepted to JCDL 2024 Research Track

With robots increasingly collaborating with humans in everyday tasks, it is important to take steps toward robotic systems capable of understanding the environment. This work focuses on scene understanding to detect pick and place tasks given initial and final images from the scene. To this end, a dataset is collected for object detection and pick and place task detection. A YOLOv5 network is subsequently trained to detect the objects in the initial and final scenes. Given the detected objects and their bounding boxes, two methods are proposed to detect the pick and place tasks which transform the initial scene into the final scene. A geometric method is proposed which tracks objects' movements in the two scenes and works based on the intersection of the bounding boxes which moved within scenes. Contrarily, the CNN-based method utilizes a Convolutional Neural Network to classify objects with intersected bounding boxes into 5 classes, showing the spatial relationship between the involved objects. The performed pick and place tasks are then derived from analyzing the experiments with both scenes. Results show that the CNN-based method, using a VGG16 backbone, outscores the geometric method by roughly 12 percentage points in certain scenarios, with an overall success rate of 84.3%., Comment: Conference Paper, ICEE 2024, 7 pages, 5 figures

Open Radio Access Network (RAN) is a transformative paradigm that supports openness, interoperability, and intelligence, with the O-RAN architecture being the most recognized framework in academia and industry. In the context of Open RAN, the importance of Energy Saving (ES) is heightened, especially with the current direction of network densification in sixth generation of mobile networks (6G). Traditional energy-saving methods in RAN struggle with the increasing dynamics of the network. This paper proposes using Reinforcement Learning (RL), a subset of Machine Learning (ML), to improve ES. We present a novel deep RL method for ES in 6G O-RAN, implemented as xApp (ES-xApp). We developed two Deep Q-Network (DQN)-based ES-xApps. ES-xApp-1 uses RSS and User Equipment (UE) geolocations, while ES-xApp-2 uses only RSS. The proposed models significantly outperformed heuristic and baseline xApps, especially with over 20 UEs. With 50 UEs, 50% of Radio Cards (RCs) were switched off, compared to 17% with the heuristic algorithm. We have observed that more informative inputs may lead to more stable training and results. This paper highlights the necessity of energy conservation in wireless networks and offers practical strategies and evidence for future research and industry practices., Comment: accepted in CAMAD 2024

This paper reviews the Challenge on Video Saliency Prediction at AIM 2024. The goal of the participants was to develop a method for predicting accurate saliency maps for the provided set of video sequences. Saliency maps are widely exploited in various applications, including video compression, quality assessment, visual perception studies, the advertising industry, etc. For this competition, a previously unused large-scale audio-visual mouse saliency (AViMoS) dataset of 1500 videos with more than 70 observers per video was collected using crowdsourced mouse tracking. The dataset collection methodology has been validated using conventional eye-tracking data and has shown high consistency. Over 30 teams registered in the challenge, and there are 7 teams that submitted the results in the final phase. The final phase solutions were tested and ranked by commonly used quality metrics on a private test subset. The results of this evaluation and the descriptions of the solutions are presented in this report. All data, including the private test subset, is made publicly available on the challenge homepage - https://challenges.videoprocessing.ai/challenges/video-saliency-prediction.html., Comment: ECCVW 2024

This paper presents a framework for human-robot collaboration that integrates optimal trajectory generation with a robust tracking control strategy. The proposed framework leverages the inverse differential Riccati equation to optimize the collaboration dynamics, providing an efficient method to generate time-varying, task-specific trajectories for the human-robot system. To ensure the accurate tracking of these trajectories, a neuro-adaptive PID control method is implemented, capable of compensating for system uncertainties and variations. This control strategy dynamically adjusts the PID gains using a radial basis function neural network, ensuring both stability and adaptability. Simulations demonstrate the method's effectiveness in achieving optimized human-robot collaboration and accurate joint-space tracking, making it suitable for real-world applications., Comment: 7 pages, 6 figures, 1 table

Though there is much interest in fair AI systems, the problem of fairness noncompliance -- which concerns whether fair models are used in practice -- has received lesser attention. Zero-Knowledge Proofs of Fairness (ZKPoF) address fairness noncompliance by allowing a service provider to verify to external parties that their model serves diverse demographics equitably, with guaranteed confidentiality over proprietary model parameters and data. They have great potential for building public trust and effective AI regulation, but no previous techniques for ZKPoF are fit for real-world deployment. We present OATH, the first ZKPoF framework that is (i) deployably efficient with client-facing communication comparable to in-the-clear ML as a Service query answering, and an offline audit phase that verifies an asymptotically constant quantity of answered queries, (ii) deployably flexible with modularity for any score-based classifier given a zero-knowledge proof of correct inference, (iii) deployably secure with an end-to-end security model that guarantees confidentiality and fairness across training, inference, and audits. We show that OATH obtains strong robustness against malicious adversaries at concretely efficient parameter settings. Notably, OATH provides a 1343x improvement to runtime over previous work for neural network ZKPoF, and scales up to much larger models -- even DNNs with tens of millions of parameters.

In the medical domain, acquiring large datasets poses significant challenges due to privacy concerns. Nonetheless, the development of a robust deep-learning model for retinal disease diagnosis necessitates a substantial dataset for training. The capacity to generalize effectively on smaller datasets remains a persistent challenge. The scarcity of data presents a significant barrier to the practical implementation of scalable medical AI solutions. To address this issue, we've combined a wide range of data sources to improve performance and generalization to new data by giving it a deeper understanding of the data representation from multi-modal datasets and developed a self-supervised framework based on large language models (LLMs), SwinV2 to gain a deeper understanding of multi-modal dataset representations, enhancing the model's ability to extrapolate to new data for the detection of eye diseases using optical coherence tomography (OCT) images. We adopt a two-phase training methodology, self-supervised pre-training, and fine-tuning on a downstream supervised classifier. An ablation study conducted across three datasets employing various encoder backbones, without data fusion, with low data availability setting, and without self-supervised pre-training scenarios, highlights the robustness of our method. Our findings demonstrate consistent performance across these diverse conditions, showcasing superior generalization capabilities compared to the baseline model, ResNet-50., Comment: 25 pages, 9 tables, 10 figures

The sum of the absolute values of the Fourier coefficients of a function $f:\mathbb{F}_2^n \to \mathbb{R}$ is called the spectral norm of $f$. Green and Sanders' quantitative version of Cohen's idempotent theorem states that if the spectral norm of $f:\mathbb{F}_2^n \to \{0,1\}$ is at most $M$, then the support of $f$ belongs to the ring of sets generated by at most $\ell(M)$ cosets, where $\ell(M)$ is a constant that only depends on $M$. We prove that the above statement can be generalized to \emph{approximate} spectral norms if and only if the support of $f$ and its complement satisfy a certain arithmetic connectivity condition. In particular, our theorem provides a new proof of the quantitative Cohen's theorem for $\mathbb{F}_2^n$.

We show that on a $\sigma$-finite measure preserving system $X = (X,\nu, T)$, the non-conventional ergodic averages $$ \mathbb{E}_{n \in [N]} \Lambda(n) f(T^n x) g(T^{P(n)} x)$$ converge pointwise almost everywhere for $f \in L^{p_1}(X)$, $g \in L^{p_2}(X)$, and $1/p_1 + 1/p_2 \leq 1$, where $P$ is a polynomial with integer coefficients of degree at least $2$. This had previously been established with the von Mangoldt weight $\Lambda$ replaced by the constant weight $1$ by the first and third authors with Mirek, and by the M\"obius weight $\mu$ by the fourth author. The proof is based on combining tools from both of these papers, together with several Gowers norm and polynomial averaging operator estimates on approximants to the von Mangoldt function of ''Cram\'er'' and ''Heath-Brown'' type., Comment: 37 pages

This paper outlines the LLMs4OL 2024, the first edition of the Large Language Models for Ontology Learning Challenge. LLMs4OL is a community development initiative collocated with the 23rd International Semantic Web Conference (ISWC) to explore the potential of Large Language Models (LLMs) in Ontology Learning (OL), a vital process for enhancing the web with structured knowledge to improve interoperability. By leveraging LLMs, the challenge aims to advance understanding and innovation in OL, aligning with the goals of the Semantic Web to create a more intelligent and user-friendly web. In this paper, we give an overview of the 2024 edition of the LLMs4OL challenge and summarize the contributions., Comment: 15 pages, 1 figure, Will appear in "The 1st LLMs4OL Challenge @ ISWC 2024" proceedings

We present Nebula, a system for differential private histogram estimation of data distributed among clients. Nebula enables clients to locally subsample and encode their data such that an untrusted server learns only data values that meet an aggregation threshold to satisfy differential privacy guarantees. Compared with other private histogram estimation systems, Nebula uniquely achieves all of the following: \textit{i)} a strict upper bound on privacy leakage; \textit{ii)} client privacy under realistic trust assumptions; \textit{iii)} significantly better utility compared to standard local differential privacy systems; and \textit{iv)} avoiding trusted third-parties, multi-party computation, or trusted hardware. We provide both a formal evaluation of Nebula's privacy, utility and efficiency guarantees, along with an empirical evaluation on three real-world datasets. We demonstrate that clients can encode and upload their data efficiently (only 0.0058 seconds running time and 0.0027 MB data communication) and privately (strong differential privacy guarantees $\varepsilon=1$). On the United States Census dataset, the Nebula's untrusted aggregation server estimates histograms with above 88\% better utility than the existing local deployment of differential privacy. Additionally, we describe a variant that allows clients to submit multi-dimensional data, with similar privacy, utility, and performance. Finally, we provide an open source implementation of Nebula.

Privacy-preserving methods for personalizing large language models (LLMs) are relatively under-explored. There are two schools of thought on this topic: (1) generating personalized outputs by personalizing the input prompt through retrieval augmentation from the user's personal information (RAG-based methods), and (2) parameter-efficient fine-tuning of LLMs per user that considers efficiency and space limitations (PEFT-based methods). This paper presents the first systematic comparison between two approaches on a wide range of personalization tasks using seven diverse datasets. Our results indicate that RAG-based and PEFT-based personalization methods on average yield 14.92% and 1.07% improvements over the non-personalized LLM, respectively. We find that combining RAG with PEFT elevates these improvements to 15.98%. Additionally, we identify a positive correlation between the amount of user data and PEFT's effectiveness, indicating that RAG is a better choice for cold-start users (i.e., user's with limited personal data).

This article considers the challenge of accommodating outlier measurements in state estimation. The Risk-Averse Performance-Specified (RAPS) state estimation approach addresses outliers as a measurement selection Bayesian risk minimization problem subject to an information accuracy constraint, which is a non-convex optimization problem. Prior explorations into RAPS rely on exhaustive search, which becomes computationally infeasible as the number of measurements increases. This paper derives a convex formulation for the RAPS optimization problems via transforming the mixed-binary variables into linear constraints. The convex reformulation herein can be solved by convex programming toolboxes, significantly enhancing computational efficiency. We explore two specifications: Full-RAPS, utilizing the full information matrix, and Diag-RAPS, focusing on diagonal elements only. The simulation comparison demonstrates that Diag-RAPS is faster and more efficient than Full-RAPS. In comparison with Kalman Filter (KF) and Threshold Decisions (TD), Diag-RAPS consistently achieves the lowest risk, while achieving the performance specification when it is feasible., Comment: Accepted by the 2024 IEEE Conference on Decision and Control

This paper aims to investigate the impact of interference in social network algorithms via user-bot interactions, focusing on the Stochastic Bounded Confidence Model (SBCM). This paper explores two approaches: positioning bots controlled by agents into the network and targeted advertising under various circumstances, operating with an advertising budget. This study integrates the Deep Deterministic Policy Gradient (DDPG) algorithm and its variants to experiment with different Deep Reinforcement Learning (DRL). Finally, experimental results demonstrate that this approach can result in efficient opinion shaping, indicating its potential in deploying advertising resources on social platforms., Comment: 5 pages, 3 figures, 2 tables

Detecting non-classical light is a central requirement for photonics-based quantum technologies. Unrivaled high efficiencies and low dark counts have positioned superconducting nanowire single photon detectors (SNSPDs) as the leading detector technology for fiber and integrated photonic applications. However, a central challenge lies in their integration within photonic integrated circuits regardless of material platform or surface topography. Here, we introduce a method based on transfer printing that overcomes these constraints and allows for the integration of SNSPDs onto arbitrary photonic substrates. We prove this by integrating SNSPDs and showing through-waveguide single-photon detection in commercially manufactured silicon and lithium niobate on insulator integrated photonic circuits. Our method eliminates bottlenecks to the integration of high-quality single-photon detectors, turning them into a versatile and accessible building block for scalable quantum information processing., Comment: 11 pages, 12 figures

$B$-Matching is a special case of matching problems where nodes can join multiple matchings with the degree of each node constrained by an upper bound, the node's $B$-value. The core solution of a bipartite $B$-matching is both a matching between the nodes respecting the upper bound constraint and an allocation of the weights of the edges among the nodes such that no group of nodes can deviate and collectively gain higher allocation. We present two learning dynamics that converge to the core of the bipartite $B$-matching problems. The first dynamics are centralized dynamics in the nature of the Hungarian method, which converge to the core in a polynomial time. The second dynamics are distributed dynamics, which converge to the core with probability one. For the distributed dynamics, a node maintains only a state consisting of (i) the aspiration levels for all of its possible matches and (ii) the matches, if any, to which it belongs. The node does not keep track of its history nor is it aware of the environment state. In each stage, a randomly activated node proposes to form a new match and changes its aspiration based on the success or failure of its proposal. At this stage, the proposing node inquires about the aspiration of the node it wants to match with to calculate the feasibility of the match. The environment matching structure changes whenever a proposal succeeds. A state is absorbing for the distributed dynamics if and only if it is in the core of the $B$-matching., Comment: 10 pages, 5 figures; accepted for CDC 2024

Prior Membership Inference Attacks (MIAs) on pre-trained Large Language Models (LLMs), adapted from classification model attacks, fail due to ignoring the generative process of LLMs across token sequences. In this paper, we present a novel attack that adapts MIA statistical tests to the perplexity dynamics of subsequences within a data point. Our method significantly outperforms prior loss-based approaches, revealing context-dependent memorization patterns in pre-trained LLMs.

We demonstrate the ability of large language models (LLMs) to perform material and molecular property regression tasks, a significant deviation from the conventional LLM use case. We benchmark the Large Language Model Meta AI (LLaMA) 3 on several molecular properties in the QM9 dataset and 24 materials properties. Only composition-based input strings are used as the model input and we fine tune on only the generative loss. We broadly find that LLaMA 3, when fine-tuned using the SMILES representation of molecules, provides useful regression results which can rival standard materials property prediction models like random forest or fully connected neural networks on the QM9 dataset. Not surprisingly, LLaMA 3 errors are 5-10x higher than those of the state-of-the-art models that were trained using far more granular representation of molecules (e.g., atom types and their coordinates) for the same task. Interestingly, LLaMA 3 provides improved predictions compared to GPT-3.5 and GPT-4o. This work highlights the versatility of LLMs, suggesting that LLM-like generative models can potentially transcend their traditional applications to tackle complex physical phenomena, thus paving the way for future research and applications in chemistry, materials science and other scientific domains.

Federated Learning (FL) is a promising distributed learning framework designed for privacy-aware applications. FL trains models on client devices without sharing the client's data and generates a global model on a server by aggregating model updates. Traditional FL approaches risk exposing sensitive client data when plain model updates are transmitted to the server, making them vulnerable to security threats such as model inversion attacks where the server can infer the client's original training data from monitoring the changes of the trained model in different rounds. Google's Secure Aggregation (SecAgg) protocol addresses this threat by employing a double-masking technique, secret sharing, and cryptography computations in honest-but-curious and adversarial scenarios with client dropouts. However, in scenarios without the presence of an active adversary, the computational and communication cost of SecAgg significantly increases by growing the number of clients. To address this issue, in this paper, we propose ACCESS-FL, a communication-and-computation-efficient secure aggregation method designed for honest-but-curious scenarios in stable FL networks with a limited rate of client dropout. ACCESS-FL reduces the computation/communication cost to a constant level (independent of the network size) by generating shared secrets between only two clients and eliminating the need for double masking, secret sharing, and cryptography computations. To evaluate the performance of ACCESS-FL, we conduct experiments using the MNIST, FMNIST, and CIFAR datasets to verify the performance of our proposed method. The evaluation results demonstrate that our proposed method significantly reduces computation and communication overhead compared to state-of-the-art methods, SecAgg and SecAgg+.

Video Anomaly Detection (VAD) presents a significant challenge in computer vision, particularly due to the unpredictable and infrequent nature of anomalous events, coupled with the diverse and dynamic environments in which they occur. Human-centric VAD, a specialized area within this domain, faces additional complexities, including variations in human behavior, potential biases in data, and substantial privacy concerns related to human subjects. These issues complicate the development of models that are both robust and generalizable. To address these challenges, recent advancements have focused on pose-based VAD, which leverages human pose as a high-level feature to mitigate privacy concerns, reduce appearance biases, and minimize background interference. In this paper, we introduce PoseWatch, a novel transformer-based architecture designed specifically for human-centric pose-based VAD. PoseWatch features an innovative Spatio-Temporal Pose and Relative Pose (ST-PRP) tokenization method that enhances the representation of human motion over time, which is also beneficial for broader human behavior analysis tasks. The architecture's core, a Unified Encoder Twin Decoders (UETD) transformer, significantly improves the detection of anomalous behaviors in video data. Extensive evaluations across multiple benchmark datasets demonstrate that PoseWatch consistently outperforms existing methods, establishing a new state-of-the-art in pose-based VAD. This work not only demonstrates the efficacy of PoseWatch but also highlights the potential of integrating Natural Language Processing techniques with computer vision to advance human behavior analysis.

PHEVA, a Privacy-preserving Human-centric Ethical Video Anomaly detection dataset. By removing pixel information and providing only de-identified human annotations, PHEVA safeguards personally identifiable information. The dataset includes seven indoor/outdoor scenes, featuring one novel, context-specific camera, and offers over 5x the pose-annotated frames compared to the largest previous dataset. This study benchmarks state-of-the-art methods on PHEVA using a comprehensive set of metrics, including the 10% Error Rate (10ER), a metric used for anomaly detection for the first time providing insights relevant to real-world deployment. As the first of its kind, PHEVA bridges the gap between conventional training and real-world deployment by introducing continual learning benchmarks, with models outperforming traditional methods in 82.14% of cases. The dataset is publicly available at https://github.com/TeCSAR-UNCC/PHEVA.git.

This paper investigates the impact of internet centralization on DNS provisioning, particularly its effects on vulnerable populations such as the indigenous people of Australia. We analyze the DNS dependencies of Australian government domains that serve indigenous communities compared to those serving the general population. Our study categorizes DNS providers into leading (hyperscaler, US-headquartered companies), non-leading (smaller Australian-headquartered or non-Australian companies), and Australian government-hosted providers. Then, we build dependency graphs to demonstrate the direct dependency between Australian government domains and their DNS providers and the indirect dependency involving further layers of providers. Additionally, we conduct an IP location analysis of DNS providers to map out the geographical distribution of DNS servers, revealing the extent of centralization on DNS services within or outside of Australia. Finally, we introduce an attacker model to categorize potential cyber attackers based on their intentions and resources. By considering attacker models and DNS dependency results, we discuss the security vulnerability of each population group against any group of attackers and analyze whether the current setup of the DNS services of Australian government services contributes to a digital divide., Comment: 12 pages, 13 figures, 5 tables Submitted to IEEE Transactions on Network and Service Management

This paper explores machine learning (ML) models for classifying lung cancer levels to improve diagnostic accuracy and prognosis. Through parameter tuning and rigorous evaluation, we assess various ML algorithms. Techniques like minimum child weight and learning rate monitoring were used to reduce overfitting and optimize performance. Our findings highlight the robust performance of Deep Neural Network (DNN) models across all phases. Ensemble methods, including voting and bagging, also showed promise in enhancing predictive accuracy and robustness. However, Support Vector Machine (SVM) models with the Sigmoid kernel faced challenges, indicating a need for further refinement. Overall, our study provides insights into ML-based lung cancer classification, emphasizing the importance of parameter tuning to optimize model performance and improve diagnostic accuracy in oncological care.

Large Language Models (LLMs) have emerged as promising solutions for a variety of medical and clinical decision support applications. However, LLMs are often subject to different types of biases, which can lead to unfair treatment of individuals, worsening health disparities, and reducing trust in AI-augmented medical tools. Aiming to address this important issue, in this study, we present a new model alignment approach for aligning LLMs using a preference optimization method within a knowledge distillation framework. Prior to presenting our proposed method, we first use an evaluation framework to conduct a comprehensive (largest to our knowledge) empirical evaluation to reveal the type and nature of existing biases in LLMs used for medical applications. We then offer a bias mitigation technique to reduce the unfair patterns in LLM outputs across different subgroups identified by the protected attributes. We show that our mitigation method is effective in significantly reducing observed biased patterns. Our code is publicly available at \url{https://github.com/healthylaife/FairAlignmentLLM}., Comment: arXiv admin note: substantial text overlap with arXiv:2404.15149

It has been a long-standing challenge to define a canonical loop integrand for non-supersymmetric gluon scattering amplitudes in the planar limit. Naive integrands are inflicted with $1/0$ ambiguities associated with tadpoles and massless external bubbles, which destroy integrand-level gauge invariance as well as consistent on-shell factorization on single loop-cuts. In this letter, we show that this essentially kinematical obstruction to defining "the" integrand for Yang-Mills theory has a structural solution, handed to us by the formulation of gluon amplitudes in terms of curves on surfaces. This defines "surface kinematics" generalizing momenta, making it possible to define "the" integrand satisfying both a (surface generalized) notion of gauge-invariance and consistent loop-cuts. The integrand also vanishes at infinity in appropriate directions, allowing it to be recursively computed for non-supersymmetric Yang-Mills theory in any number of dimensions. We illustrate these ideas through one loop for all multiplicity, and for the simplest two-loop integrand., Comment: 13 pages, 4 figures, explicit expressions at tree-level and one-loop given in supplementary materials

Large Language Models (LLMs) have seen increasing use in various software development tasks, especially in code generation. The most advanced recent methods attempt to incorporate feedback from code execution into prompts to help guide LLMs in generating correct code, in an iterative process. While effective, these methods could be costly and time-consuming due to numerous interactions with the LLM and the extensive token usage. To address this issue, we propose an alternative approach named Evolutionary Prompt Engineering for Code (EPiC), which leverages a lightweight evolutionary algorithm to evolve the original prompts toward better ones that produce high-quality code, with minimal interactions with LLM. Our evaluation against state-of-the-art (SOTA) LLM-based code generation models shows that EPiC outperforms all the baselines in terms of cost-effectiveness., Comment: Submitted to TSE

We present the first measurement of the user-effect and privacy impact of "Related Website Sets," a recent proposal to reduce browser privacy protections between two sites if those sites are related to each other. An assumption (both explicitly and implicitly) underpinning the Related Website Sets proposal is that users can accurately determine if two sites are related via the same entity. In this work, we probe this assumption via measurements and a user study of 30 participants, to assess the ability of Web users to determine if two sites are (according to the Related Website Sets feature) related to each other. We find that this is largely not the case. Our findings indicate that 42 (36.8%) of the user determinations in our study are incorrect in privacy-harming ways, where users think that sites are not related, but would be treated as related (and so due less privacy protections) by the Related Website Sets feature. Additionally, 22 (73.3%) of participants made at least one incorrect evaluation during the study. We also characterise the Related Website Sets list, its composition over time, and its governance.