Search

We introduce BLIP3-KALE, a dataset of 218 million image-text pairs that bridges the gap between descriptive synthetic captions and factual web-scale alt-text. KALE augments synthetic dense image captions with web-scale alt-text to generate factually grounded image captions. Our two-stage approach leverages large vision-language models and language models to create knowledge-augmented captions, which are then used to train a specialized VLM for scaling up the dataset. We train vision-language models on KALE and demonstrate improvements on vision-language tasks. Our experiments show the utility of KALE for training more capable and knowledgeable multimodal models. We release the KALE dataset at https://huggingface.co/datasets/Salesforce/blip3-kale

Underground forums serve as hubs for cybercriminal activities, offering a space for anonymity and evasion of conventional online oversight. In these hidden communities, malicious actors collaborate to exchange illicit knowledge, tools, and tactics, driving a range of cyber threats from hacking techniques to the sale of stolen data, malware, and zero-day exploits. Identifying the key instigators (i.e., key hackers), behind these operations is essential but remains a complex challenge. This paper presents a novel method called EUREKHA (Enhancing User Representation for Key Hacker Identification in Underground Forums), designed to identify these key hackers by modeling each user as a textual sequence. This sequence is processed through a large language model (LLM) for domain-specific adaptation, with LLMs acting as feature extractors. These extracted features are then fed into a Graph Neural Network (GNN) to model user structural relationships, significantly improving identification accuracy. Furthermore, we employ BERTopic (Bidirectional Encoder Representations from Transformers Topic Modeling) to extract personalized topics from user-generated content, enabling multiple textual representations per user and optimizing the selection of the most representative sequence. Our study demonstrates that fine-tuned LLMs outperform state-of-the-art methods in identifying key hackers. Additionally, when combined with GNNs, our model achieves significant improvements, resulting in approximately 6% and 10% increases in accuracy and F1-score, respectively, over existing methods. EUREKHA was tested on the Hack-Forums dataset, and we provide open-source access to our code., Comment: Accepted at IEEE Trustcom 2024

An accurate description of elastic effects of coherent microstructures is necessary for the predictive modeling of microstructural evolution in many structural materials. To date, there has not been a demonstration on how continuum elasticity models are able to reproduce finite-temperature stress-fields and elastic energy estimates of coherent precipitates from atomistic simulations. We present a comparison of stress-fields of coherent precipitates in the body-centered cubic (BCC) Fe-Cr system obtained from atomistic simulation data and from continuum elasticity modeling. The magnitude and topology of the stress-fields show good agreement between the two approaches, and we show the importance of elastic effects on the Gibbs-Thompson effect for this small lattice misfit system. We conclude with a discussion of potential complications of continuum modeling for systems with larger misfit., Comment: 19 pages, 5 figures, 4 tables, supplemental material in anc directory in source

Treatment selection in breast cancer is guided by molecular subtypes and clinical characteristics. Recurrence risk assessment plays a crucial role in personalizing treatment. Current methods, including genomic assays, have limited accuracy and clinical utility, leading to suboptimal decisions for many patients. We developed a test for breast cancer patient stratification based on digital pathology and clinical characteristics using novel AI methods. Specifically, we utilized a vision transformer-based pan-cancer foundation model trained with self-supervised learning to extract features from digitized H&E-stained slides. These features were integrated with clinical data to form a multi-modal AI test predicting cancer recurrence and death. The test was developed and evaluated using data from a total of 8,161 breast cancer patients across 15 cohorts originating from seven countries. Of these, 3,502 patients from five cohorts were used exclusively for evaluation, while the remaining patients were used for training. Our test accurately predicted our primary endpoint, disease-free interval, in the five external cohorts (C-index: 0.71 [0.68-0.75], HR: 3.63 [3.02-4.37, p<0.01]). In a direct comparison (N=858), the AI test was more accurate than Oncotype DX, the standard-of-care 21-gene assay, with a C-index of 0.67 [0.61-0.74] versus 0.61 [0.49-0.73], respectively. Additionally, the AI test added independent information to Oncotype DX in a multivariate analysis (HR: 3.11 [1.91-5.09, p<0.01)]). The test demonstrated robust accuracy across all major breast cancer subtypes, including TNBC (C-index: 0.71 [0.62-0.81], HR: 3.81 [2.35-6.17, p=0.02]), where no diagnostic tools are currently recommended by clinical guidelines. These results suggest that our AI test can improve accuracy, extend applicability to a wider range of patients, and enhance access to treatment selection tools.

Mosques are worship places of Allah and must be preserved clean, immaculate, provide all the comforts of the worshippers in them. The prophet's mosque in Medina/ Saudi Arabia is one of the most important mosques for Muslims. It occupies second place after the sacred mosque in Mecca/ Saudi Arabia, which is in constant overcrowding by all Muslims to visit the prophet Mohammad's tomb. This paper aims to propose a smart dome model to preserve the fresh air and allow the sunlight to enter the mosque using artificial intelligence techniques. The proposed model controls domes movements based on the weather conditions and the overcrowding rates in the mosque. The data have been collected from two different resources, the first one from the database of Saudi Arabia weather's history, and the other from Shanghai Technology Database. Congested Scene Recognition Network (CSRNet) and Fuzzy techniques have applied using Python programming language to control the domes to be opened and closed for a specific time to renew the air inside the mosque. Also, this model consists of several parts that are connected for controlling the mechanism of opening/closing domes according to weather data and the situation of crowding in the mosque. Finally, the main goal of this paper has been achieved, and the proposed model has worked efficiently and specifies the exact duration time to keep the domes open automatically for a few minutes for each hour head.

Reinforcement learning with general utilities has recently gained attention thanks to its ability to unify several problems, including imitation learning, pure exploration, and safe RL. However, prior work for solving this general problem in a unified way has mainly focused on the tabular setting. This is restrictive when considering larger state-action spaces because of the need to estimate occupancy measures during policy optimization. In this work, we address this issue and propose to approximate occupancy measures within a function approximation class using maximum likelihood estimation (MLE). We propose a simple policy gradient algorithm (PG-OMA) where an actor updates the policy parameters to maximize the general utility objective whereas a critic approximates the occupancy measure using MLE. We provide a sample complexity analysis of PG-OMA showing that our occupancy measure estimation error only scales with the dimension of our function approximation class rather than the size of the state action space. Under suitable assumptions, we establish first order stationarity and global optimality performance bounds for the proposed PG-OMA algorithm for nonconcave and concave general utilities respectively. We complement our methodological and theoretical findings with promising empirical results showing the scalability potential of our approach compared to existing tabular count-based approaches., Comment: 26 pages, 5 figures

The widely used expected utility theory has been shown to be empirically inconsistent with human preferences in the psychology and behavioral economy literatures. Cumulative Prospect Theory (CPT) has been developed to fill in this gap and provide a better model for human-based decision-making supported by empirical evidence. It allows to express a wide range of attitudes and perceptions towards risk, gains and losses. A few years ago, CPT has been combined with Reinforcement Learning (RL) to formulate a CPT policy optimization problem where the goal of the agent is to search for a policy generating long-term returns which are aligned with their preferences. In this work, we revisit this policy optimization problem and provide new insights on optimal policies and their nature depending on the utility function under consideration. We further derive a novel policy gradient theorem for the CPT policy optimization objective generalizing the seminal corresponding result in standard RL. This result enables us to design a model-free policy gradient algorithm to solve the CPT-RL problem. We illustrate the performance of our algorithm in simple examples motivated by traffic control and electricity management applications. We also demonstrate that our policy gradient algorithm scales better to larger state spaces compared to the existing zeroth order algorithm for solving the same problem., Comment: 33 pages, 19 figures

Key generation, a pillar in physical-layer security (PLS), is the process of the exchanging signals from two legitimate users (Alice and Bob) to extract a common key from the random, common channels. The drawback of extracting keys from wireless channels is the ample dependence on the dynamicity and fluctuations of the radio channel, rendering the key vulnerable to estimation by Eve (an illegitimate user) in low-entropy environments because of insufficient randomness. Added to that, the lack of channel fluctuations lower the secret key rate (SKR) defined as the number of bits of key generated per channel use. In this work, we aim to address this challenge by using a reconfigurable intelligent surface (RIS) to produce random phases at certain, carefully curated intervals such that it disrupts the channel in low-entropy environments. We propose an RIS assisted key generation protocol, study its performance, and compare with benchmarks to observe the benefit of using an RIS while considering various important metrics such as key mismatch rate and secret key throughput. Furthermore, we characterize a scaling law as a function of the rate of change of RIS phase switching for the average secret information rate under this protocol. Then, we use both the key throughput and information rate to optimize the overall secrecy rate. Simulations are made to validate our theoretical findings and effectiveness of the proposed scheme showing an improvement in performance when an RIS is deployed., Comment: arXiv admin note: text overlap with arXiv:2308.04726

Artificial neural networks (ANNs) represent a fundamentally connectionnist and distributed approach to computing, and as such they differ from classical computers that utilize the von Neumann architecture. This has revived research interest in new unconventional hardware to enable more efficient implementations of ANNs rather than emulating them on traditional machines. In order to fully leverage the capabilities of this new generation of ANNs, optimization algorithms that take into account hardware limitations and imperfections are necessary. Photonics represents a particularly promising platform, offering scalability, high speed, energy efficiency, and the capability for parallel information processing. Yet, fully fledged implementations of autonomous optical neural networks (ONNs) with in-situ learning remain scarce. In this work, we propose a ternary weight architecture high-dimensional semiconductor laser-based ONN. We introduce a simple method for achieving ternary weights with Boolean hardware, significantly increasing the ONN's information processing capabilities. Furthermore, we design a novel in-situ optimization algorithm that is compatible with, both, Boolean and ternary weights, and provide a detailed hyperparameter study of said algorithm for two different tasks. Our novel algorithm results in benefits, both in terms of convergence speed and performance. Finally, we experimentally characterize the long-term inference stability of our ONN and find that it is extremely stable with a consistency above 99\% over a period of more than 10 hours, addressing one of the main concerns in the field. Our work is of particular relevance in the context of in-situ learning under restricted hardware resources, especially since minimizing the power consumption of auxiliary hardware is crucial to preserving efficiency gains achieved by non-von Neumann ANN implementations.

Pre-trained on extensive and diverse multi-modal datasets, 2D foundation models excel at addressing 2D tasks with little or no downstream supervision, owing to their robust representations. The emergence of 2D-to-3D distillation frameworks has extended these capabilities to 3D models. However, distilling 3D representations for autonomous driving datasets presents challenges like self-similarity, class imbalance, and point cloud sparsity, hindering the effectiveness of contrastive distillation, especially in zero-shot learning contexts. Whereas other methodologies, such as similarity-based distillation, enhance zero-shot performance, they tend to yield less discriminative representations, diminishing few-shot performance. We investigate the gap in structure between the 2D and the 3D representations that result from state-of-the-art distillation frameworks and reveal a significant mismatch between the two. Additionally, we demonstrate that the observed structural gap is negatively correlated with the efficacy of the distilled representations on zero-shot and few-shot 3D semantic segmentation. To bridge this gap, we propose a relational distillation framework enforcing intra-modal and cross-modal constraints, resulting in distilled 3D representations that closely capture the structure of the 2D representation. This alignment significantly enhances 3D representation performance over those learned through contrastive distillation in zero-shot segmentation tasks. Furthermore, our relational loss consistently improves the quality of 3D representations in both in-distribution and out-of-distribution few-shot segmentation tasks, outperforming approaches that rely on the similarity loss., Comment: Accepted at ECCV 2024

Background and Objective: Diabetes presents a significant challenge to healthcare due to the negative impact of poor blood sugar control on health and associated complications. Computer simulation platforms, notably exemplified by the UVA/Padova Type 1 Diabetes simulator, has emerged as a promising tool for advancing diabetes treatments by simulating patient responses in a virtual environment. The UVA Virtual Lab (UVLab) is a new simulation platform to mimic the metabolic behavior of people with Type 2 diabetes (T2D) with a large population of 6062 virtual subjects. Methods: The work introduces the Distribution-Based Population Selection (DSPS) method, a systematic approach to identifying virtual subsets that mimic the clinical behavior observed in real trials. The method transforms the sub-population selection task into a Linear Programing problem, enabling the identification of the largest representative virtual cohort. This selection process centers on key clinical outcomes in diabetes research, such as HbA1c and Fasting plasma Glucose (FPG), ensuring that the statistical properties (moments) of the selected virtual sub-population closely resemble those observed in real-word clinical trial. Results: DSPS method was applied to the insulin degludec (IDeg) arm of a phase 3 clinical trial. This method was used to select a sub-population of virtual subjects that closely mirrored the clinical trial data across multiple key metrics, including glycemic efficacy, insulin dosages, and cumulative hypoglycemia events over a 26-week period. Conclusion: The DSPS algorithm is able to select virtual sub-population within UVLab to reproduce and predict the outcomes of a clinical trial. This statistical method can bridge the gap between large population simulation platforms and previously conducted clinical trials.

The increasing demand for medical imaging has surpassed the capacity of available radiologists, leading to diagnostic delays and potential misdiagnoses. Artificial intelligence (AI) techniques, particularly in automatic medical report generation (AMRG), offer a promising solution to this dilemma. This review comprehensively examines AMRG methods from 2021 to 2024. It (i) presents solutions to primary challenges in this field, (ii) explores AMRG applications across various imaging modalities, (iii) introduces publicly available datasets, (iv) outlines evaluation metrics, (v) identifies techniques that significantly enhance model performance, and (vi) discusses unresolved issues and potential future research directions. This paper aims to provide a comprehensive understanding of the existing literature and inspire valuable future research., Comment: 42 pages and 9 figures

We present xGen-VideoSyn-1, a text-to-video (T2V) generation model capable of producing realistic scenes from textual descriptions. Building on recent advancements, such as OpenAI's Sora, we explore the latent diffusion model (LDM) architecture and introduce a video variational autoencoder (VidVAE). VidVAE compresses video data both spatially and temporally, significantly reducing the length of visual tokens and the computational demands associated with generating long-sequence videos. To further address the computational costs, we propose a divide-and-merge strategy that maintains temporal consistency across video segments. Our Diffusion Transformer (DiT) model incorporates spatial and temporal self-attention layers, enabling robust generalization across different timeframes and aspect ratios. We have devised a data processing pipeline from the very beginning and collected over 13M high-quality video-text pairs. The pipeline includes multiple steps such as clipping, text detection, motion estimation, aesthetics scoring, and dense captioning based on our in-house video-LLM model. Training the VidVAE and DiT models required approximately 40 and 642 H100 days, respectively. Our model supports over 14-second 720p video generation in an end-to-end way and demonstrates competitive performance against state-of-the-art T2V models., Comment: Accepted by ECCV24 AI4VA

Machine vision enhances automation, quality control, and operational efficiency in industrial applications by enabling machines to interpret and act on visual data. While traditional computer vision algorithms and approaches remain widely utilized, machine learning has become pivotal in current research activities. In particular, generative AI demonstrates promising potential by improving pattern recognition capabilities, through data augmentation, increasing image resolution, and identifying anomalies for quality control. However, the application of generative AI in machine vision is still in its early stages due to challenges in data diversity, computational requirements, and the necessity for robust validation methods. A comprehensive literature review is essential to understand the current state of generative AI in industrial machine vision, focusing on recent advancements, applications, and research trends. Thus, a literature review based on the PRISMA guidelines was conducted, analyzing over 1,200 papers on generative AI in industrial machine vision. Our findings reveal various patterns in current research, with the primary use of generative AI being data augmentation, for machine vision tasks such as classification and object detection. Furthermore, we gather a collection of application challenges together with data requirements to enable a successful application of generative AI in industrial machine vision. This overview aims to provide researchers with insights into the different areas and applications within current research, highlighting significant advancements and identifying opportunities for future work., Comment: 44 pages, 7 figures, This work has been submitted to the Journal of Intelligent Manufacturing

This paper introduces cuVegas, a CUDA-based implementation of the Vegas Enhanced Algorithm (VEGAS+), optimized for multi-dimensional integration in GPU environments. The VEGAS+ algorithm is an advanced form of Monte Carlo integration, recognized for its adaptability and effectiveness in handling complex, high-dimensional integrands. It employs a combination of variance reduction techniques, namely adaptive importance sampling and a variant of adaptive stratified sampling, that make it particularly adept at managing integrands with multiple peaks or those aligned with the diagonals of the integration volume. Being a Monte Carlo integration method, the task is well suited for parallelization and for GPU execution. Our implementation, cuVegas, aims to harness the inherent parallelism of GPUs, addressing the challenge of workload distribution that often hampers efficiency in standard implementations. We present a comprehensive analysis comparing cuVegas with existing CPU and GPU implementations, demonstrating significant performance improvements, from two to three orders of magnitude on CPUs, and from a factor of two on GPUs over the best existing implementation. We also demonstrate the speedup for integrands for which VEGAS+ was designed, with multiple peaks or other significant structures aligned with diagonals of the integration volume., Comment: 25 pages, 8 figures

This report introduces xGen-MM (also known as BLIP-3), a framework for developing Large Multimodal Models (LMMs). The framework comprises meticulously curated datasets, a training recipe, model architectures, and a resulting suite of LMMs. xGen-MM, short for xGen-MultiModal, expands the Salesforce xGen initiative on foundation AI models. Our models undergo rigorous evaluation across a range of tasks, including both single and multi-image benchmarks. Our pre-trained base model exhibits strong in-context learning capabilities and the instruction-tuned model demonstrates competitive performance among open-source LMMs with similar model sizes. In addition, we introduce a safety-tuned model with DPO, aiming to mitigate harmful behaviors such as hallucinations and improve safety. We open-source our models, curated large-scale datasets, and our fine-tuning codebase to facilitate further advancements in LMM research. Associated resources will be available on our project page above.

Markov Potential Games (MPGs) form an important sub-class of Markov games, which are a common framework to model multi-agent reinforcement learning problems. In particular, MPGs include as a special case the identical-interest setting where all the agents share the same reward function. Scaling the performance of Nash equilibrium learning algorithms to a large number of agents is crucial for multi-agent systems. To address this important challenge, we focus on the independent learning setting where agents can only have access to their local information to update their own policy. In prior work on MPGs, the iteration complexity for obtaining $\epsilon$-Nash regret scales linearly with the number of agents $N$. In this work, we investigate the iteration complexity of an independent policy mirror descent (PMD) algorithm for MPGs. We show that PMD with KL regularization, also known as natural policy gradient, enjoys a better $\sqrt{N}$ dependence on the number of agents, improving over PMD with Euclidean regularization and prior work. Furthermore, the iteration complexity is also independent of the sizes of the agents' action spaces., Comment: 16 pages, CDC 2024

Otoscopy is a diagnostic procedure to examine the ear canal and eardrum using an otoscope. It identifies conditions like infections, foreign bodies, ear drum perforations and ear abnormalities. We propose a novel resource efficient deep learning and transformer based framework, Sumotosima (Summarizer for otoscopic images), an end-to-end pipeline for classification followed by summarization. Our framework works on combination of triplet and cross-entropy losses. Additionally, we use Knowledge Enhanced Multimodal BART whose input is fused textual and image embedding. The objective is to provide summaries that are well-suited for patients, ensuring clarity and efficiency in understanding otoscopic images. Given the lack of existing datasets, we have curated our own OCASD (Otoscopic Classification And Summary Dataset), which includes 500 images with 5 unique categories annotated with their class and summaries by Otolaryngologists. Sumotosima achieved a result of 98.03%, which is 7.00%, 3.10%, 3.01% higher than K-Nearest Neighbors, Random Forest and Support Vector Machines, respectively, in classification tasks. For summarization, Sumotosima outperformed GPT-4o and LLaVA by 88.53% and 107.57% in ROUGE scores, respectively. We have made our code and dataset publicly available at https://github.com/anas2908/Sumotosima, Comment: Work in Progress

Reliably manufacturing defect free products is still an open challenge for Laser Powder Bed Fusion processes. Particularly, pores that occur frequently have a negative impact on mechanical properties like fatigue performance. Therefore, an accurate localisation of pores is mandatory for quality assurance, but requires time-consuming post-processing steps like computer tomography scans. Although existing solutions using in-situ monitoring data can detect pore occurrence within a layer, they are limited in their localisation precision. Therefore, we propose a pore localisation approach that estimates their position within a single layer using a Gaussian kernel density estimation. This allows segmentation models to learn the correlation between in-situ monitoring data and the derived probability distribution of pore occurrence. Within our experiments, we compare the prediction performance of different segmentation models depending on machine parameter configuration and geometry features. From our results, we conclude that our approach allows a precise localisation of pores that requires minimal data preprocessing. Our research extends the literature by providing a foundation for more precise pore detection systems., Comment: 20 pages, 7 figures, This work has been submitted to the Journal Progress in Additive Manufacturing

We present the Code Documentation and Analysis Tool (CoDAT). CoDAT is a tool designed to maintain consistency between the various levels of code documentation, e.g. if a line in a code sketch is changed, the comment that documents the corresponding code is also changed. That is, comments are linked and updated so as to remain internally consistent and also consistent with the code. By flagging "out of date" comments, CoDAT alerts the developer to maintain up-to-date documentation. We use a large language model to check the semantic consistency between a fragment of code and the comments that describe it. Thus we also flag semantic inconsistency as well as out of date comments. This helps programers write code that correctly implements a code sketch, and so provides machine support for a step-wise refinement approach, starting with a code sketch and proceeding down to code through one or more refinement iterations. CoDAT is implemented in the Intellij IDEA IDE where we use the Code Insight daemon package alongside a custom regular expression algorithm to mark tagged comments whose corresponding code blocks have changed. CoDAT's backend is structurally decentralized to allow a distributed ledger framework for code consistency and architectural compilation tracking., Comment: 31 pages

Inspired by piecewise polynomiality results of double Hurwitz numbers, Ardila and Brugall\'e introduced an enumerative problem which they call double Gromov--Witten invariants of Hirzebruch surfaces. These invariants serve as a two-dimensional analogue and satisfy a similar piecewise polynomial structure. More precisely, they introduced the enumeration of curves in Hirzebruch surfaces satisfying point conditions and tangency conditions on the two parallel toric boundaries. These conditions are stored in four partitions and the resulting invariants are piecewise polynomial in their entries. Moreover, they found that these expressions also behave polynomially with respect to the parameter determining the underlying Hirzebruch surfaces. Based on work of Ardila and Block, they proposed that such a polynomiality could also hold while changing between more general toric surfaces corresponding to $h$-transverse polygons. In this work, we answer this question affirmatively. Moreover, we express the resulting invariants for $h$-transverse polygons as matrix elements in the two-dimensional bosonic Fock space., Comment: 24 pages, 8 figures, 3 tables. arXiv admin note: text overlap with arXiv:1412.4563 by other authors

Generative Flow Networks (GFlowNets) have emerged as an innovative learning paradigm designed to address the challenge of sampling from an unnormalized probability distribution, called the reward function. This framework learns a policy on a constructed graph, which enables sampling from an approximation of the target probability distribution through successive steps of sampling from the learned policy. To achieve this, GFlowNets can be trained with various objectives, each of which can lead to the model s ultimate goal. The aspirational strength of GFlowNets lies in their potential to discern intricate patterns within the reward function and their capacity to generalize effectively to novel, unseen parts of the reward function. This paper attempts to formalize generalization in the context of GFlowNets, to link generalization with stability, and also to design experiments that assess the capacity of these models to uncover unseen parts of the reward function. The experiments will focus on length generalization meaning generalization to states that can be constructed only by longer trajectories than those seen in training.

The ability to acknowledge the inevitable uncertainty in their knowledge and reasoning is a prerequisite for AI systems to be truly truthful and reliable. In this paper, we present a taxonomy of uncertainty specific to vision-language AI systems, distinguishing between epistemic uncertainty (arising from a lack of information) and aleatoric uncertainty (due to inherent unpredictability), and further explore finer categories within. Based on this taxonomy, we synthesize a benchmark dataset, CertainlyUncertain, featuring 178K visual question answering (VQA) samples as contrastive pairs. This is achieved by 1) inpainting images to make previously answerable questions into unanswerable ones; and 2) using image captions to prompt large language models for both answerable and unanswerable questions. Additionally, we introduce a new metric confidence-weighted accuracy, that is well correlated with both accuracy and calibration error, to address the shortcomings of existing metrics., Comment: 26 pages

The relaxation of atomic positions to their optimal structural arrangement is crucial for understanding the emergence of new physical behavior in long scale superstructures in twisted bilayers of two-dimensional materials. The amount of deviation from a rigid moir\'e structure will depend on the elastic properties of the constituent monolayers which for the twisted bilayer - the more flexible the monolayers are, the lower the energy required to deform the layers to maximize the areas with an energetically optimal interlayer arrangement of atoms. We investigate this atomic reconstruction for twisted bilayers of highly flexible InSe. Results using two methods are demonstrated - first we train a machine-learned interatomic potential (MLIP) to enable fully atomistic relaxations of small-twist-angle large-length-scale moir\'e supercells while retaining density functional theory (DFT) level accuracy. We find substantial out-of-plane corrugation and in-plane domain formation for a wide range of twist angles and moir\'e length scales. We then adapt an existing continuum approach and show that it can reproduce some, but not all, features of the fully atomistic calculations., Comment: 11 pages

Calculating mealtime insulin doses poses a significant challenge for individuals with Type 1 Diabetes (T1D). Doses should perfectly compensate for expected post-meal glucose excursions, requiring a profound understanding of the individual's insulin sensitivity and the meal macronutrients'. Usually, people rely on intuition and experience to develop this understanding. In this work, we demonstrate how a reinforcement learning agent, employing a self-attention encoder network, can effectively mimic and enhance this intuitive process. Trained on 80 virtual subjects from the FDA-approved UVA/Padova T1D adult cohort and tested on twenty, self-attention demonstrates superior performance compared to other network architectures. Results reveal a significant reduction in glycemic risk, from 16.5 to 9.6 in scenarios using sensor-augmented pump and from 9.1 to 6.7 in scenarios using automated insulin delivery. This new paradigm bypasses conventional therapy parameters, offering the potential to simplify treatment and promising improved quality of life and glycemic outcomes for people with T1D., Comment: 6 pages, 4 figures, Biological and Medical Systems - 12th BMS 2024 - IFAC

In this manuscript, we introduce a tensor-based approach to Non-Negative Tensor Factorization (NTF). The method entails tensor dimension reduction through the utilization of the Einstein product. To maintain the regularity and sparsity of the data, certain constraints are imposed. Additionally, we present an optimization algorithm in the form of a tensor multiplicative updates method, which relies on the Einstein product. To guarantee a minimum number of iterations for the convergence of the proposed algorithm, we employ the Reduced Rank Extrapolation (RRE) and the Topological Extrapolation Transformation Algorithm (TEA). The efficacy of the proposed model is demonstrated through tests conducted on Hyperspectral Images (HI) for denoising, as well as for Hyperspectral Image Linear Unmixing. Numerical experiments are provided to substantiate the effectiveness of the proposed model for both synthetic and real data.

Multimodal interleaved datasets featuring free-form interleaved sequences of images and text are crucial for training frontier large multimodal models (LMMs). Despite the rapid progression of open-source LMMs, there remains a pronounced scarcity of large-scale, diverse open-source multimodal interleaved datasets. In response, we introduce MINT-1T, the most extensive and diverse open-source Multimodal INTerleaved dataset to date. MINT-1T comprises one trillion text tokens and 3.4 billion images, a 10x scale-up from existing open-source datasets. Additionally, we include previously untapped sources such as PDFs and ArXiv papers. As scaling multimodal interleaved datasets requires substantial engineering effort, sharing the data curation process and releasing the dataset greatly benefits the community. Our experiments show that LMMs trained on MINT-1T rival the performance of models trained on the previous leading dataset, OBELICS. Our data and code will be released at https://github.com/mlfoundations/MINT-1T.

Physical neural networks (PNNs) are a class of neural-like networks that leverage the properties of physical systems to perform computation. While PNNs are so far a niche research area with small-scale laboratory demonstrations, they are arguably one of the most underappreciated important opportunities in modern AI. Could we train AI models 1000x larger than current ones? Could we do this and also have them perform inference locally and privately on edge devices, such as smartphones or sensors? Research over the past few years has shown that the answer to all these questions is likely "yes, with enough research": PNNs could one day radically change what is possible and practical for AI systems. To do this will however require rethinking both how AI models work, and how they are trained - primarily by considering the problems through the constraints of the underlying hardware physics. To train PNNs at large scale, many methods including backpropagation-based and backpropagation-free approaches are now being explored. These methods have various trade-offs, and so far no method has been shown to scale to the same scale and performance as the backpropagation algorithm widely used in deep learning today. However, this is rapidly changing, and a diverse ecosystem of training techniques provides clues for how PNNs may one day be utilized to create both more efficient realizations of current-scale AI models, and to enable unprecedented-scale models., Comment: 29 pages, 4 figures

Suffix trees are key and efficient data structure for solving string problems. A suffix tree is a compressed trie containing all the suffixes of a given text of length $n$ with a linear construction cost. In this work, we introduce an algorithm to build a linear index that allows finding a pattern of any length under any internal node in a suffix tree in O(logn) time.

Visualizations, such as charts, are crucial for interpreting complex data. However, they are often provided as raster images, which are not compatible with assistive technologies for people with blindness and visual impairments, such as embossed papers or tactile displays. At the same time, creating accessible vector graphics requires a skilled sighted person and is time-intensive. In this work, we leverage advancements in the field of chart analysis to generate tactile charts in an end-to-end manner. Our three key contributions are as follows: (1) introducing the ChartFormer model trained to convert raster chart images into tactile-accessible SVGs, (2) training this model on the Chart2Tactile dataset, a synthetic chart dataset we created following accessibility standards, and (3) evaluating the effectiveness of our SVGs through a pilot user study with an refreshable two-dimensional tactile display. Our work is publicly available at https://github.com/nsothman/ChartFormer ., Comment: Accepted at ICCHP 2024. Codes will be available at https://github.com/nsothman/ChartFormer

Following the recent popularity of Large Language Models (LLMs), several attempts have been made to extend them to the visual domain. From having a visual assistant that could guide us through unfamiliar environments to generative models that produce images using only a high-level text description, the vision-language model (VLM) applications will significantly impact our relationship with technology. However, there are many challenges that need to be addressed to improve the reliability of those models. While language is discrete, vision evolves in a much higher dimensional space in which concepts cannot always be easily discretized. To better understand the mechanics behind mapping vision to language, we present this introduction to VLMs which we hope will help anyone who would like to enter the field. First, we introduce what VLMs are, how they work, and how to train them. Then, we present and discuss approaches to evaluate VLMs. Although this work primarily focuses on mapping images to language, we also discuss extending VLMs to videos.

Spike-based encoders represent information as sequences of spikes or pulses, which are transmitted between neurons. A prevailing consensus suggests that spike-based approaches demonstrate exceptional capabilities in capturing the temporal dynamics of neural activity and have the potential to provide energy-efficient solutions for low-power applications. The Spiketrum encoder efficiently compresses input data using spike trains or code sets (for non-spiking applications) and is adaptable to both hardware and software implementations, with lossless signal reconstruction capability. The paper proposes and assesses Spiketrum's hardware, evaluating its output under varying spike rates and its classification performance with popular spiking and non-spiking classifiers, and also assessing the quality of information compression and hardware resource utilization. The paper extensively benchmarks both Spiketrum hardware and its software counterpart against state-of-the-art, biologically-plausible encoders. The evaluations encompass benchmarking criteria, including classification accuracy, training speed, and sparsity when using encoder outputs in pattern recognition and classification with both spiking and non-spiking classifiers. Additionally, they consider encoded output entropy and hardware resource utilization and power consumption of the hardware version of the encoders. Results demonstrate Spiketrum's superiority in most benchmarking criteria, making it a promising choice for various applications. It efficiently utilizes hardware resources with low power consumption, achieving high classification accuracy. This work also emphasizes the potential of encoders in spike-based processing to improve the efficiency and performance of neural computing systems., Comment: To be published at "IEEE/ACM Transactions on Audio, Speech, and Language Processing"

This paper presents a novel FPGA-based neuromorphic cochlea, leveraging the general-purpose spike-coding algorithm, Spiketrum. The focus of this study is on the development and characterization of this cochlea model, which excels in transforming audio vibrations into biologically realistic auditory spike trains. These spike trains are designed to withstand neural fluctuations and spike losses while accurately encapsulating the spatial and precise temporal characteristics of audio, along with the intensity of incoming vibrations. Noteworthy features include the ability to generate real-time spike trains with minimal information loss and the capacity to reconstruct original signals. This fine-tuning capability allows users to optimize spike rates, achieving an optimal balance between output quality and power consumption. Furthermore, the integration of a feedback system into Spiketrum enables selective amplification of specific features while attenuating others, facilitating adaptive power consumption based on application requirements. The hardware implementation supports both spike-based and non-spike-based processors, making it versatile for various computing systems. The cochlea's ability to encode diverse sensory information, extending beyond sound waveforms, positions it as a promising sensory input for current and future spike-based intelligent computing systems, offering compact and real-time spike train generation., Comment: To be published at "IEEE Transactions on Circuits and Systems"

Accurate approximation of complex nonlinear functions is a fundamental challenge across many scientific and engineering domains. Traditional neural network architectures, such as Multi-Layer Perceptrons (MLPs), often struggle to efficiently capture intricate patterns and irregularities present in high-dimensional functions. This paper presents the Chebyshev Kolmogorov-Arnold Network (Chebyshev KAN), a new neural network architecture inspired by the Kolmogorov-Arnold representation theorem, incorporating the powerful approximation capabilities of Chebyshev polynomials. By utilizing learnable functions parametrized by Chebyshev polynomials on the network's edges, Chebyshev KANs enhance flexibility, efficiency, and interpretability in function approximation tasks. We demonstrate the efficacy of Chebyshev KANs through experiments on digit classification, synthetic function approximation, and fractal function generation, highlighting their superiority over traditional MLPs in terms of parameter efficiency and interpretability. Our comprehensive evaluation, including ablation studies, confirms the potential of Chebyshev KANs to address longstanding challenges in nonlinear function approximation, paving the way for further advancements in various scientific and engineering applications.

Offline scheduling in Time Sensitive Networking (TSN) utilizing the Time Aware Shaper (TAS) facilitates optimal deterministic latency and jitter-bounds calculation for Time- Triggered (TT) flows. However, the dynamic nature of traffic in industrial settings necessitates a strategy for adaptively scheduling flows without interrupting existing schedules. Our research identifies critical gaps in current dynamic scheduling methods for TSN and introduces the novel GCN-TD3 approach. This novel approach utilizes a Graph Convolutional Network (GCN) for representing the various relations within different components of TSN and employs the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm to dynamically schedule any incoming flow. Additionally, an Integer Linear Programming (ILP) based offline scheduler is used both to initiate the simulation and serve as a fallback mechanism. This mechanism is triggered to recalculate the entire schedule when the predefined threshold of Gate Control List(GCL) length exceeds. Comparative analyses demonstrate that GCN-TD3 outperforms existing methods like Deep Double Q-Network (DDQN) and Deep Deterministic Policy Gradient (DDPG), exhibiting convergence within 4000 epochs with a 90\% dynamic TT flow admission rate while maintaining deadlines and reducing jitter to as low as 2us. Finally, two modules were developed for the OMNeT++ simulator, facilitating dynamic simulation to evaluate the methodology., Comment: Accepted article IFIP/IEEE Networking 2024 (Tensor)

The Black-Litterman model is a framework for incorporating forward-looking expert views in a portfolio optimization problem. Existing work focuses almost exclusively on single-period problems and assumes that the horizon of expert forecasts matches that of the investor. We consider a multi-period generalization where the horizon of expert views may differ from that of a dynamically-trading investor. By exploiting an underlying graphical structure relating the asset prices and views, we derive the conditional distribution of asset returns when the price process is geometric Brownian motion. We also show that it can be written in terms of a multi-dimensional Brownian bridge. The new price process is an affine factor model with the conditional log-price process playing the role of a vector of factors. We derive an explicit expression for the optimal dynamic investment policy and analyze the hedging demand associated with the new covariate. More generally, the paper shows that Bayesian graphical models are a natural framework for incorporating complex information structures in the Black-Litterman model.

Managing the semantic quality of the categorization in large textual datasets, such as Wikipedia, presents significant challenges in terms of complexity and cost. In this paper, we propose leveraging transformer models to distill semantic information from texts in the Wikipedia dataset and its associated categories into a latent space. We then explore different approaches based on these encodings to assess and enhance the semantic identity of the categories. Our graphical approach is powered by Convex Hull, while we utilize Hierarchical Navigable Small Worlds (HNSWs) for the hierarchical approach. As a solution to the information loss caused by the dimensionality reduction, we modulate the following mathematical solution: an exponential decay function driven by the Euclidean distances between the high-dimensional encodings of the textual categories. This function represents a filter built around a contextual category and retrieves items with a certain Reconsideration Probability (RP). Retrieving high-RP items serves as a tool for database administrators to improve data groupings by providing recommendations and identifying outliers within a contextual framework.

We present LayerSkip, an end-to-end solution to speed-up inference of large language models (LLMs). First, during training we apply layer dropout, with low dropout rates for earlier layers and higher dropout rates for later layers, and an early exit loss where all transformer layers share the same exit. Second, during inference, we show that this training recipe increases the accuracy of early exit at earlier layers, without adding any auxiliary layers or modules to the model. Third, we present a novel self-speculative decoding solution where we exit at early layers and verify and correct with remaining layers of the model. Our proposed self-speculative decoding approach has less memory footprint than other speculative decoding approaches and benefits from shared compute and activations of the draft and verification stages. We run experiments on different Llama model sizes on different types of training: pretraining from scratch, continual pretraining, finetuning on specific data domain, and finetuning on specific task. We implement our inference solution and show speedups of up to 2.16x on summarization for CNN/DM documents, 1.82x on coding, and 2.0x on TOPv2 semantic parsing task. We open source our code and checkpoints at https://github.com/facebookresearch/LayerSkip., Comment: ACL 2024

We present a discrete filter for subgrid-scale (SGS) model, coupled with the discretization corrected particle strength exchange (DC-PSE) method for the simulation of three-dimensional viscous incompressible flow, at high Reynolds flows. The majority of turbulence modelling techniques, particularly in complex geometries, face significant computational challenges due to the difficulties in implementing 3-D convolution operations for asymmetric boundary conditions or curved domain boundaries. In this contribution Taylor expansion is used to define differential operators corresponding to the convolution filter, so that the transfer function remains very close to the unity of sizeable displacement in wave number, making the filter a good approximation to the convolution one. A discrete Gaussian filter, in both fourth and second-order forms, was evaluated with varying ratios of particle spacing to the cut-off length. The impact of the filter's order and the ratio's value is thoroughly examined and detailed in the study. Additionally, the Brinkman penalisation technique is employed to impose boundary conditions implicitly, allowing for efficient and accurate flow simulations around complex geometries without the need for modifying the numerical method or computational domain. The incompressible flow is governed by the the he Entropically Damped Artificial Compressibility equations allowing explicit simulation of the incompressible Navier-Stokes equations. The effectiveness of the proposed methodology is validated through several benchmark problems, including isotropic turbulence decay, and flow around four cylinders arranged in a square in-line configuration. These test cases demonstrate the method's accuracy in capturing the intricate flow structures characteristic of high Reynolds number flows, highlighting its applicability to industrial turbulence modeling.

Foundation models are a strong trend in deep learning and computer vision. These models serve as a base for applications as they require minor or no further fine-tuning by developers to integrate into their applications. Foundation models for zero-shot object segmentation such as Segment Anything (SAM) output segmentation masks from images without any further object information. When they are followed in a pipeline by an object identification model, they can perform object detection without training. Here, we focus on training such an object identification model. A crucial practical aspect for an object identification model is to be flexible in input size. As object identification is an image retrieval problem, a suitable method should handle multi-query multi-gallery situations without constraining the number of input images (e.g. by having fixed-size aggregation layers). The key solution to train such a model is the centroid triplet loss (CTL), which aggregates image features to their centroids. CTL yields high accuracy, avoids misleading training signals and keeps the model input size flexible. In our experiments, we establish a new state of the art on the ArmBench object identification task, which shows general applicability of our model. We furthermore demonstrate an integrated unseen object detection pipeline on the challenging HOPE dataset, which requires fine-grained detection. There, our pipeline matches and surpasses related methods which have been trained on dataset-specific data., Comment: Accepted to CASE 2024

In this paper, we analyze the effectiveness of transfer learning on classifying electronic components. Transfer learning reuses pre-trained models to save time and resources in building a robust classifier rather than learning from scratch. Our work introduces a lightweight CNN, coined as VoltaVision, and compares its performance against more complex models. We test the hypothesis that transferring knowledge from a similar task to our target domain yields better results than state-of-the-art models trained on general datasets. Our dataset and code for this work are available at https://github.com/AnasIshfaque/VoltaVision., Comment: Tiny Paper at ICLR 2024

Unsupervised (US) video anomaly detection (VAD) in surveillance applications is gaining more popularity recently due to its practical real-world applications. As surveillance videos are privacy sensitive and the availability of large-scale video data may enable better US-VAD systems, collaborative learning can be highly rewarding in this setting. However, due to the extremely challenging nature of the US-VAD task, where learning is carried out without any annotations, privacy-preserving collaborative learning of US-VAD systems has not been studied yet. In this paper, we propose a new baseline for anomaly detection capable of localizing anomalous events in complex surveillance videos in a fully unsupervised fashion without any labels on a privacy-preserving participant-based distributed training configuration. Additionally, we propose three new evaluation protocols to benchmark anomaly detection approaches on various scenarios of collaborations and data availability. Based on these protocols, we modify existing VAD datasets to extensively evaluate our approach as well as existing US SOTA methods on two large-scale datasets including UCF-Crime and XD-Violence. All proposed evaluation protocols, dataset splits, and codes are available here: https://github.com/AnasEmad11/CLAP, Comment: Accepted in IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR), 2024