Search

AI language models (LMs) show promise for biological sequence analysis. We re-engineered open-source LMs (GPT-2, BLOOM, DistilRoBERTa, ELECTRA, and Mamba, ranging from 70M to 12B parameters) for microbial sequence classification. The models achieved F1 scores up to 95 and operated 16,580x faster and at 2.9x the recall of BLASTP. They effectively classified the algal dark proteome - uncharacterized proteins comprising about 65% of total proteins - validated on new data including a new, complete Hi-C/Pacbio Chlamydomonas genome. Larger (>1B) LA4SR models reached high accuracy (F1 > 86) when trained on less than 2% of available data, rapidly achieving strong generalization capacity. High accuracy was achieved when training data had intact or scrambled terminal information, demonstrating robust generalization to incomplete sequences. Finally, we provide custom AI explainability software tools for attributing amino acid patterns to AI generative processes and interpret their outputs in evolutionary and biophysical contexts.

Many workers in cities across the world, who have been teleworking because of the COVID-19 pandemic, are expected to be back to their commutes. As this process is believed to be gradual and telecommuting is likely to remain an option for many workers, hybrid model and flexible schedules might become the norm in the future. This variable work schedules allows employees to commute outside of traditional rush hours. Moreover, many studies showed that commuters might be skeptical of using trains, buses, and carpools and could turn to personal vehicles to get to work, which might increase congestion and crashes in the roads. This study attempts to provide information on the safest time to commute to Washington, DC area analyzing historical traffic crash data before the COVID-19 pandemic. It also aims to advance our understanding of traffic crashes and other relating factors such as weather in the Washington, DC area. We created a model to predict crashes by time of the day, using a negative binomial regression after rejecting a Poisson regression, and additionally explored the validity of a Random Forest regression. Our main consideration for an eventual application of this study is to reduce crashes in Washington DC, using this tool that provides people with better options on when to commute and when to telework, if available. The study also provides policymakers and researchers with real-world insights that decrease the number of traffic crashes to help achieve the goals of The Vision Zero Initiative adopted by the district.

Band-selective absorption and emission of thermal radiation in the infrared are of interest due to applications in emissivity coatings, infrared sensing, thermo-photovoltaics and solar energy harvesting. The broadband nature of thermal radiation presents distinct challenges in achieving spectral and angular selectivity, which are difficult to address by prevalent optical strategies, often yielding restrictive responses. We explore a tri-layer coating employing a nanostructured grating of epsilon-near-zero (ENZ) material, indium tin oxide (ITO), atop a dielectric (silicon dioxide) and metal (gold) underlayer, which shows wide-angle (0-60 degrees) and band-selective (1800 - 2800 nm) high absorption (> 0.8). Numerical simulations and experimental results reveal that the ENZ response of ITO combined with its localized plasmon resonances define the high absorption bandwidth, aided by the sandwiched dielectric's optical properties, elucidating the tunability of the absorption bandwidth. Thermal imaging in the mid-infrared highlights the relevance of the ENZ grating, emphasizing the potential of this coating design as a thermal emitter. This study offers valuable insights into light-matter interactions and opens avenues for practical applications in thermal management and energy harvesting.

In this study, 0.5m high resolution satellite datasets over Indian urban region was used to demonstrate the applicability of deep learning models over Ahmedabad, India. Here, YOLOv7 instance segmentation model was trained on well curated trees canopy dataset (6500 images) in order to carry out the change detection. During training, evaluation metrics such as bounding box regression and mask regression loss, mean average precision (mAP) and stochastic gradient descent algorithm were used for evaluating and optimizing the performance of model. After the 500 epochs, the mAP of 0.715 and 0.699 for individual tree detection and tree canopy mask segmentation were obtained. However, by further tuning hyper parameters of the model, maximum accuracy of 80 % of trees detection with false segmentation rate of 2% on data was obtained.

Automatic Speech Recognition (ASR) systems have been examined and shown to exhibit biases toward particular groups of individuals, influenced by factors such as demographic traits, accents, and speech styles. Noise can disproportionately impact speakers with certain accents, dialects, or speaking styles, leading to biased error rates. In this work, we introduce a novel framework DENOASR, which is a selective denoising technique to reduce the disparity in the word error rates between the two gender groups, male and female. We find that a combination of two popular speech denoising techniques, viz. DEMUCS and LE, can be effectively used to mitigate ASR disparity without compromising their overall performance. Experiments using two state-of-the-art open-source ASRs - OpenAI WHISPER and NVIDIA NEMO - on multiple benchmark datasets, including TIE, VOX-POPULI, TEDLIUM, and FLEURS, show that there is a promising reduction in the average word error rate gap across the two gender groups. For a given dataset, the denoising is selectively applied on speech samples having speech intelligibility below a certain threshold, estimated using a small validation sample, thus ameliorating the need for large-scale human-written ground-truth transcripts. Our findings suggest that selective denoising can be an elegant approach to mitigate biases in present-day ASR systems., Comment: Paper accepted at IEEE ICKG 2024

Despite the significant influx of prompt-tuning techniques for generative vision-language models (VLMs), it remains unclear how sensitive these models are to lexical and semantic alterations in prompts. In this paper, we evaluate the ability of generative VLMs to understand lexical and semantic changes in text using the SugarCrepe++ dataset. We analyze the sensitivity of VLMs to lexical alterations in prompts without corresponding semantic changes. Our findings demonstrate that generative VLMs are highly sensitive to such alterations. Additionally, we show that this vulnerability affects the performance of techniques aimed at achieving consistency in their outputs.

The vast amount of data generated by camera sensors has prompted the exploration of energy-efficient processing solutions for deploying computer vision tasks on edge devices. Among the various approaches studied, processing-in-pixel integrates massively parallel analog computational capabilities at the extreme-edge, i.e., within the pixel array and exhibits enhanced energy and bandwidth efficiency by generating the output activations of the first neural network layer rather than the raw sensory data. In this article, we propose an energy and bandwidth efficient ADC-less processing-in-pixel architecture. This architecture implements an optimized binary activation neural network trained using Hoyer regularizer for high accuracy on complex vision tasks. In addition, we also introduce a global shutter burst memory read scheme utilizing fast and disturb-free read operation leveraging innovative use of nanoscale voltage-controlled magnetic tunnel junctions (VC-MTJs). Moreover, we develop an algorithmic framework incorporating device and circuit constraints (characteristic device switching behavior and circuit non-linearity) based on state-of-the-art fabricated VC-MTJ characteristics and extensive circuit simulations using commercial GlobalFoundries 22nm FDX technology. Finally, we evaluate the proposed system's performance on two complex datasets - CIFAR10 and ImageNet, showing improvements in front-end and communication energy efficiency by 8.2x and 8.5x respectively and reduction in bandwidth by 6x compared to traditional computer vision systems, without any significant drop in the test accuracy., Comment: 25 pages, 9 figures, 1 table

Network pruning has emerged as a potential solution to make LLMs cheaper to deploy. However, existing LLM pruning approaches universally rely on the C4 dataset as the calibration data for calculating pruning scores, leaving its optimality unexplored. In this study, we evaluate the choice of calibration data on LLM pruning, across a wide range of datasets that are most commonly used in LLM training and evaluation, including four pertaining datasets as well as three categories of downstream tasks encompassing nine datasets. Each downstream dataset is prompted with In-Context Learning (ICL) and Chain-of-Thought (CoT), respectively. Besides the already intriguing observation that the choice of calibration data significantly impacts the performance of pruned LLMs, our results also uncover several subtle and often unexpected findings, summarized as follows: (1) C4 is not the optimal choice for LLM pruning, even among commonly used pre-training datasets; (2) arithmetic datasets, when used as calibration data, performs on par or even better than pre-training datasets; (3) pruning with downstream datasets does not necessarily help the corresponding downstream task, compared to pre-training data; (4) ICL is widely beneficial to all data categories, whereas CoT is only useful on certain tasks. Our findings shed light on the importance of carefully selecting calibration data for LLM pruning and pave the way for more efficient deployment of these powerful models in real-world applications. We release our code at: https://github.com/abx393/llm-pruning-calibration-data., Comment: EMNLP 2024

The dynamical and geometric structures of the Broad Line Region (BLR), along with the origins of continuum time delays in active galaxies, remain topics of ongoing debate. In this study, we aim to reproduce the broadband spectrum, the H$\beta$ line delay, and the continuum time delays available for the source NGC 5548. We employ the standard accretion disk model with the option of an inner hot flow, alongside the lamp-post model to account for disk irradiation and a BLR structure model based on radiation pressure acting on dust. The model is parameterized by the black hole mass (fixed), accretion rate, viewing angle, lamp-post height, cloud density, and cloud covering factor. The resulting continuum delays are calculated as a combination of disk reprocessing and the reprocessing of a fraction of radiation by the BLR. Our model reasonably reproduces the observed broad-band continuum, the H$\beta$ delay, and the continuum inter-band time delays measured during the observational campaign. When the accretion rate is not fixed based on the known distance to the source, we can directly estimate the distance from our model. The resulting value of H$_0$ = $69.03^{+17.81}_{-11.75}$ km s$^{-1}$ Mpc$^{-1}$ represents a noteworthy improvement compared to the findings of Cackett et al. (2007). This pilot study demonstrates that, with sufficient data coverage, it is possible to disentangle the time delays originating from the accretion disk and the BLR. This paves the way for effectively using inter-band continuum time delays in determining the Hubble constant. Additionally, the findings support the adopted model for the formation of the H$\beta$ line., Comment: submitted to A&A, comments welcome

Equal-time commutators of different components of the energy-momentum tensor at spatially separated points are calculated for a relativistic quantum Fermi gas at finite temperature and density. Different definitions of such components, also known as different pseudogauges, are used and smeared with a Gaussian profile characterized by the width $\sigma$. In this way, we introduce observables that may represent measurements of energy and momentum in a spatial region of size $\sigma$. We find that the obtained commutators are sensitive to the pseudogauge chosen if the probed systems or the spatial separation are small. The pseudogauge dependence is expected as different quantum operators are analyzed in this case. On the other hand, we find that for sufficiently large probed systems or with large separation, the studied commutators are pseudogauge independent., Comment: 22 pages, 3 figures

Estimating the location and intensity of tropical cyclones holds crucial significance for predicting catastrophic weather events. In this study, we approach this task as a detection and regression challenge, specifically over the North Indian Ocean (NIO) region where best tracks location and wind speed information serve as the labels. The current process for cyclone detection and intensity estimation involves physics-based simulation studies which are time-consuming, only using image features will automate the process for significantly faster and more accurate predictions. While conventional methods typically necessitate substantial prior knowledge for training, we are exploring alternative approaches to enhance efficiency. This research aims to focus specifically on cyclone detection, intensity estimation and related aspects using only image input and data-driven approaches and will lead to faster inference time and automate the process as opposed to current NWP models being utilized at SAC. In context to algorithm development, a novel two stage detection and intensity estimation module is proposed. In the first level detection we try to localize the cyclone over an entire image as captured by INSAT3D over the NIO (North Indian Ocean). For the intensity estimation task, we propose a CNN-LSTM network, which works on the cyclone centered images, utilizing a ResNet-18 backbone, by which we are able to capture both temporal and spatial characteristics., Comment: 10 pages, 22 figures

Current video-based computer vision (CV) applications typically suffer from high energy consumption due to reading and processing all pixels in a frame, regardless of their significance. While previous works have attempted to reduce this energy by skipping input patches or pixels and using feedback from the end task to guide the skipping algorithm, the skipping is not performed during the sensor read phase. As a result, these methods can not optimize the front-end sensor energy. Moreover, they may not be suitable for real-time applications due to the long latency of modern CV networks that are deployed in the back-end. To address this challenge, this paper presents a custom-designed reconfigurable CMOS image sensor (CIS) system that improves energy efficiency by selectively skipping uneventful regions or rows within a frame during the sensor's readout phase, and the subsequent analog-to-digital conversion (ADC) phase. A novel masking algorithm intelligently directs the skipping process in real-time, optimizing both the front-end sensor and back-end neural networks for applications including autonomous driving and augmented/virtual reality (AR/VR). Our system can also operate in standard mode without skipping, depending on application needs. We evaluate our hardware-algorithm co-design framework on object detection based on BDD100K and ImageNetVID, and gaze estimation based on OpenEDS, achieving up to 53% reduction in front-end sensor energy while maintaining state-of-the-art (SOTA) accuracy., Comment: Under review

Video game playing is an extremely structured domain where algorithmic decision-making can be tested without adverse real-world consequences. While prevailing methods rely on image inputs to avoid the problem of hand-crafting state space representations, this approach systematically diverges from the way humans actually learn to play games. In this paper, we design object-based input representations that generalize well across a number of video games. Using these representations, we evaluate an agent's ability to learn games similar to an infant - with limited world experience, employing simple inductive biases derived from intuitive representations of physics from the real world. Using such biases, we construct an object category representation to be used by a Q-learning algorithm and assess how well it learns to play multiple games based on observed object affordances. Our results suggest that a human-like object interaction setup capably learns to play several video games, and demonstrates superior generalizability, particularly for unfamiliar objects. Further exploring such methods will allow machines to learn in a human-centric way, thus incorporating more human-like learning benefits., Comment: 7 pages, Accepted in Proceedings of the Annual Meeting of the Cognitive Science Society, Volume 46

Early detection of neurodegenerative disorders is an important open problem, since early diagnosis and treatment may yield a better prognosis. Researchers have recently sought to leverage advances in machine learning algorithms to detect symptoms of altered gait, possibly corresponding to the emergence of neurodegenerative etiologies. However, while several claims of positive and accurate detection have been made in the recent literature, using a variety of sensors and algorithms, solutions are far from being realized in practice. This paper analyzes existing approaches to identify gaps inhibiting translation. Using a set of experiments across three Kinect-simulated and one real Parkinson's patient datasets, we highlight possible sources of errors and generalization failures in these approaches. Based on these observations, we propose our strong baseline called Asynchronous Multi-Stream Graph Convolutional Network (AMS-GCN) that can reliably differentiate multiple categories of pathological gaits across datasets., Comment: 23 pages, Accepted in Machine Learning for Healthcare(MLHC) 2024, JMLR Volume 252

We propose a variational approach to explore quasiparticle excitations in interacting quantum many-body systems, motivated by the potential in leveraging near-term noisy intermediate scale quantum devices for quantum state preparation. By exploiting translation invariance and potentially other abelian symmetries of the many-body Hamiltonian, we extend the variational quantum eigensolver (VQE) approach to construct spatially localized quasiparticle states that encode information on the whole excited band, allowing us to achieve quantum parallelism. We benchmark the proposed algorithm via numerical simulations performed on the one-dimension transverse field Ising chain. We show that VQE can capture both the magnon quasiparticles of the paramagnetic phase, and the topologically non-trivial domain wall excitations in the ferromagnetic regime. We show that the localized quasiparticle states constructed with VQE contain accessible information on the full band of quasiparticles, and provide valuable insight into the way interactions renormalize the bare spin flip or domain wall excitations of the simple, trivially solvable limits of the model. These results serve as important theoretical input towards utilizing quantum simulators to directly access the quasiparticles of strongly interacting quantum systems, as well as to gain insight into crucial experimentally measured properties directly determined by the nature of these quasiparticles., Comment: 16 pages, 8 figures

We examine the framework of relativistic spin-hydrodynamics in the context of electron hydrodynamics in graphene. We develop a spin-hydrodynamic model for a (2 + 1)-dimensional system of fermions under the condition of small spin polarization. Our analysis confirms that thermal vorticity, which satisfies the global equilibrium condition, is also a solution to the spin-hydrodynamic equations. Additionally, we calculate the magnetization of the system in global equilibrium and introduce a novel phenomenon - thermovortical magnetization - resulting from thermal vorticity, which can be experimentally observed in graphene., Comment: 6 pages

Training emotion recognition models has relied heavily on human annotated data, which present diversity, quality, and cost challenges. In this paper, we explore the potential of Large Language Models (LLMs), specifically GPT4, in automating or assisting emotion annotation. We compare GPT4 with supervised models and or humans in three aspects: agreement with human annotations, alignment with human perception, and impact on model training. We find that common metrics that use aggregated human annotations as ground truth can underestimate the performance, of GPT-4 and our human evaluation experiment reveals a consistent preference for GPT-4 annotations over humans across multiple datasets and evaluators. Further, we investigate the impact of using GPT-4 as an annotation filtering process to improve model training. Together, our findings highlight the great potential of LLMs in emotion annotation tasks and underscore the need for refined evaluation methodologies., Comment: to be published in Interspeech 2024

Vision-language models (VLMs) have shown impressive zero- and few-shot performance on real-world visual question answering (VQA) benchmarks, alluding to their capabilities as visual reasoning engines. However, the benchmarks being used conflate "pure" visual reasoning with world knowledge, and also have questions that involve a limited number of reasoning steps. Thus, it remains unclear whether a VLM's apparent visual reasoning performance is due to its world knowledge, or due to actual visual reasoning capabilities. To clarify this ambiguity, we systematically benchmark and dissect the zero-shot visual reasoning capabilities of VLMs through synthetic datasets that require minimal world knowledge, and allow for analysis over a broad range of reasoning steps. We focus on two novel aspects of zero-shot visual reasoning: i) evaluating the impact of conveying scene information as either visual embeddings or purely textual scene descriptions to the underlying large language model (LLM) of the VLM, and ii) comparing the effectiveness of chain-of-thought prompting to standard prompting for zero-shot visual reasoning. We find that the underlying LLMs, when provided textual scene descriptions, consistently perform better compared to being provided visual embeddings. In particular, 18% higher accuracy is achieved on the PTR dataset. We also find that CoT prompting performs marginally better than standard prompting only for the comparatively large GPT-3.5-Turbo (175B) model, and does worse for smaller-scale models. This suggests the emergence of CoT abilities for visual reasoning in LLMs at larger scales even when world knowledge is limited. Overall, we find limitations in the abilities of VLMs and LLMs for more complex visual reasoning, and highlight the important role that LLMs can play in visual reasoning., Comment: 21 pages

The ground state of a one-dimensional spin-1/2 uniform antiferromagnetic Heisenberg chain (AfHc) is a Tomonaga-Luttinger liquid which is quantum-critical with respect to applied magnetic fields upto a saturation field Hs beyond which it transforms to a fully polarised state. Wilson ratio has been predicted to be a good indicator for demarcating these phases [Phys. Rev. B 96, 220401 (2017)]. From detailed temperature and magnetic field dependent magnetisation, magnetic susceptibility and specific heat measurements in a metalorganic complex and comparisons with field theory and quantum transfer matrix method calculations, the complex was found to be a very good realisation of a spin-1/2 AfHc. Wilson ratio obtained from experimentally obtained magnetic susceptibility and magnetic contribution of specific heat values was used to map the magnetic phase diagram of the uniform spin-1/2 AfHc over large regions of phase space demarcating Tomonaga-Luttinger liquid, saturation field quantum critical, and fully polarised states. Luttinger parameter and spinon velocity were found to match very well with the values predicted from conformal field theory., Comment: Accepted for publication in PNAS Nexus

As Artificial Intelligence (AI) changes how businesses work, there is a growing need for people who can work in this sector. This paper investigates how well universities in United Kingdom offering courses in AI, prepare students for jobs in the real world. To gain insight into the differences between university curricula and industry demands we review the contents of taught courses and job advertisement portals. By using custom data scraping tools to gather information from job advertisements and university curricula, and frequency and Naive Bayes classifier analysis, this study will show exactly what skills industry is looking for. In this study we identified 12 skill categories that were used for mapping. The study showed that the university curriculum in the AI domain is well balanced in most technical skills, including Programming and Machine learning subjects, but have a gap in Data Science and Maths and Statistics skill categories., Comment: Accepted to the journal "Industry and Higher Education"

Large Language Models (LLMs) have revolutionised the capability of AI models in comprehending and generating natural language text. They are increasingly being used to empower and deploy agents in real-world scenarios, which make decisions and take actions based on their understanding of the context. Therefore researchers, policy makers and enterprises alike are working towards ensuring that the decisions made by these agents align with human values and user expectations. That being said, human values and decisions are not always straightforward to measure and are subject to different cognitive biases. There is a vast section of literature in Behavioural Science which studies biases in human judgements. In this work we report an ongoing investigation on alignment of LLMs with human judgements affected by order bias. Specifically, we focus on a famous human study which showed evidence of order effects in similarity judgements, and replicate it with various popular LLMs. We report the different settings where LLMs exhibit human-like order effect bias and discuss the implications of these findings to inform the design and development of LLM based applications., Comment: Accepted at The First Workshop on AI Behavioral Science (AIBS 2024), held in conjunction with KDD 2024

Dynamic Vision Sensors (DVS) have emerged as a revolutionary technology with a high temporal resolution that far surpasses RGB cameras. DVS technology draws biological inspiration from photoreceptors and the initial retinal synapse. Our research showcases the potential of additional retinal functionalities to extract visual features. We provide a domain-agnostic and efficient algorithm for ego-motion compensation based on Object Motion Sensitivity (OMS), one of the multiple robust features computed within the mammalian retina. We develop a framework based on experimental neuroscience that translates OMS' biological circuitry to a low-overhead algorithm. OMS processes DVS data from dynamic scenes to perform pixel-wise object motion segmentation. Using a real and a synthetic dataset, we highlight OMS' ability to differentiate object motion from ego-motion, bypassing the need for deep networks. This paper introduces a bio-inspired computer vision method that dramatically reduces the number of parameters by a factor of 1000 compared to prior works. Our work paves the way for robust, high-speed, and low-bandwidth decision-making for in-sensor computations.

We present results from the molecular dynamics (MD) simulation of surface-directed spinodal decomposition (SDSD) in binary fluid mixtures ($A+B$) with off-critical compositions. The aim is to elucidate the role of composition ratio in the early-time wetting kinetics under the influence of long-range surface potential. In our simulations, the attractive part of surface potential varies as $V(z)= -\epsilon_a/z^{n}$, with $\epsilon_{a}$ being the surface-potential strength. The surface prefers `$A$' species to form the wetting layer. Its thickness [$R_1(t)$] for the majority wetting (number of $A$-type particles [$N_A$] > number of $B$-type particles [$N_B$]), grows as a power-law with an exponent $1/(n+2)$. This is consistent with the early-time kinetics in the form of potential-dependent growth present in the Puri-Binder model. However, for minority wetting ($N_A$ < $N_B$), the growth exponent in $R_1(t)$ is less than $1/(n+2)$. Furthermore, on decreasing the field strength $\epsilon_{a}$, we recover $1/(n+2)$ for a minority wetting case. We provide phenomenological arguments to explain the early-time wetting kinetics for both cases., Comment: 11 pages, 7 figures

Human feedback on conversations with language language models (LLMs) is central to how these systems learn about the world, improve their capabilities, and are steered toward desirable and safe behaviors. However, this feedback is mostly collected by frontier AI labs and kept behind closed doors. In this work, we bring together interdisciplinary experts to assess the opportunities and challenges to realizing an open ecosystem of human feedback for AI. We first look for successful practices in peer production, open source, and citizen science communities. We then characterize the main challenges for open human feedback. For each, we survey current approaches and offer recommendations. We end by envisioning the components needed to underpin a sustainable and open human feedback ecosystem. In the center of this ecosystem are mutually beneficial feedback loops, between users and specialized models, incentivizing a diverse stakeholders community of model trainers and feedback providers to support a general open feedback pool.

We consider the problem of source sampling and transmission scheduling for age-of-information minimization in a system consisting of multiple energy harvesting (EH) sources and a sink node. At each time, one of the sources is selected by the scheduler and the quality of its channel to the sink is measured. This probed channel quality is then used to decide whether a source will sample an observation and transmit the packet to the sink in that time slot. We formulate this problem as a constrained Markov decision process (CMDP) assuming i.i.d. energy arrival and channel fading processes, and relax it using a Lagrange multiplier. We apply a near optimal Whittle's index policy to decide the node to be probed. Next, for the probed node, we derive an optimal threshold policy, which recommends source sampling and observation transmission from the probed source only when the measured channel quality is above a threshold. Our proposed policy is called Whittle's index and threshold based source scheduling and sampling (WITS3) policy. However, in order to calculate Whittle's indices, one must be aware of the underlying processes' transition matrices, which are occasionally concealed from the scheduler. Therefore, we further propose a variant Q-WITS3 of WITS3 based on Q-learning assisted by two timescale asynchronous stochastic approximation, which seeks to learn Whittle's indices and optimal policies for the case with unknown channel states and EH characteristics. Numerical results demonstrate the efficacy of our algorithms over two baseline policies., Comment: 10 pages, 7 figures

The rapid advancement of neural network applications necessitates hardware that not only accelerates computation but also adapts efficiently to dynamic processing requirements. While processing-in-pixel has emerged as a promising solution to overcome the bottlenecks of traditional architectures at the extreme-edge, existing implementations face limitations in reconfigurability and scalability due to their static nature and inefficient area usage. Addressing these challenges, we present a novel architecture that significantly enhances the capabilities of processing-in-pixel for convolutional neural networks. Our design innovatively integrates non-volatile memory (NVM) with novel unit pixel circuit design, enabling dynamic reconfiguration of synaptic weights, kernel size, channel size and stride size. Thus offering unprecedented flexibility and adaptability. With using a separate die for pixel circuit and storing synaptic weights, our circuit achieves a substantial reduction in the required area per pixel thereby increasing the density and scalability of the pixel array. Simulation results demonstrate dot product operations of the circuit, the non-linearity of its analog output and a novel bucket-select curvefit model is proposed to capture it. This work not only addresses the limitations of current in-pixel computing approaches but also opens new avenues for developing more efficient, flexible, and scalable neural network hardware, paving the way for advanced AI applications.

Recent advances in retinal neuroscience have fueled various hardware and algorithmic efforts to develop retina-inspired solutions for computer vision tasks. In this work, we focus on a fundamental visual feature within the mammalian retina, Object Motion Sensitivity (OMS). Using DVS data from EV-IMO dataset, we analyze the performance of an algorithmic implementation of OMS circuitry for motion segmentation in presence of ego-motion. This holistic analysis considers the underlying constraints arising from the hardware circuit implementation. We present novel CMOS circuits that implement OMS functionality inside image sensors, while providing run-time re-configurability for key algorithmic parameters. In-sensor technologies for dynamical environment adaptation are crucial for ensuring high system performance. Finally, we verify the functionality and re-configurability of the proposed CMOS circuit designs through Cadence simulations in 180nm technology. In summary, the presented work lays foundation for hardware-algorithm re-engineering of known biological circuits to suit application needs.

This article presents the reversible hydrogen storage capacities of Li-decorated Si6C14 and Si8C12 using Density Functional Theory (DFT). The chemical stabilities of the designed Si6C14Li6 and Si8C12Li4 nanocages are investigated using HOMO-LUMO gaps and various global reactivity descriptors such as chemical hardness and electrophilicity index. Our study reveals that each Li atom decorated over the designed Si6C14 and Si8C12 nanocages can hold up to 5H2 molecules with adsorption energy lying in the optimum range of 0.14-0.085 eV, thereby yielding an overall gravimetric density of 13.8% and 9.2% respectively. The interaction between adsorbed H2 molecules and the Li metal sites is found to occur via non-covalent and closed shell type of interaction. The H2 molecules are adsorbed in a quasi-molecular fashion with elongated bond length. The molecular dynamics study reveals that most of the H2 molecules get desorbed from the designed host nanocages at 300K without causing any significant structural changes, which confirms their reversibility. When the adsorption and desorption conditions are set at 100K/60bar and 240K/1bar respectively, the practical storage gravimetric densities of Si6C14Li6 and Si8C12Li4 cages are estimated to be 13.73 wt% and 9.08 wt%, which are relatively high in comparison to the US-DOE target of 5.5 wt% by the year 2020. Hence, the computationally designed Si6C14Li6 and Si8C12Li4 nanocages can be regarded as prospective systems for hydrogen storage applications.

Facial Recognition Systems (FRSs) are being developed and deployed globally at unprecedented rates. Most platforms are designed in a limited set of countries but deployed in worldwide, without adequate checkpoints. This is especially problematic for Global South countries which lack strong legislation to safeguard persons facing disparate performance of these systems. A combination of unavailability of datasets, lack of understanding of FRS functionality and low-resource bias mitigation measures accentuate the problem. In this work, we propose a new face dataset composed of 6,579 unique male and female sportspersons from eight countries around the world. More than 50% of the dataset comprises individuals from the Global South countries and is demographically diverse. To aid adversarial audits and robust model training, each image has four adversarial variants, totaling over 40,000 images. We also benchmark five popular FRSs, both commercial and open-source, for the task of gender prediction (and country prediction for one of the open-source models as an example of red-teaming). Experiments on industrial FRSs reveal accuracies ranging from 98.2%--38.1%, with a large disparity between males and females in the Global South (max difference of 38.5%). Biases are also observed in all FRSs between females of the Global North and South (max difference of ~50%). Grad-CAM analysis identifies the nose, forehead and mouth as the regions of interest on one of the open-source FRSs. Utilizing this insight, we design simple, low-resource bias mitigation solutions using few-shot and novel contrastive learning techniques significantly improving the accuracy with disparity between males and females reducing from 50% to 1.5% in one of the settings. In the red-teaming experiment with the open-source Deepface model, contrastive learning proves more effective than simple fine-tuning., Comment: This work has been accepted for publication at AAAI/ACM AIES 2024

Graph Neural Networks (GNNs) have attracted immense attention in the past decade due to their numerous real-world applications built around graph-structured data. On the other hand, Large Language Models (LLMs) with extensive pretrained knowledge and powerful semantic comprehension abilities have recently shown a remarkable ability to benefit applications using vision and text data. In this paper, we investigate how LLMs can be leveraged in a computationally efficient fashion to benefit rich graph-structured data, a modality relatively unexplored in LLM literature. Prior works in this area exploit LLMs to augment every node features in an ad-hoc fashion (not scalable for large graphs), use natural language to describe the complex structural information of graphs, or perform computationally expensive finetuning of LLMs in conjunction with GNNs. We propose E-LLaGNN (Efficient LLMs augmented GNNs), a framework with an on-demand LLM service that enriches message passing procedure of graph learning by enhancing a limited fraction of nodes from the graph. More specifically, E-LLaGNN relies on sampling high-quality neighborhoods using LLMs, followed by on-demand neighborhood feature enhancement using diverse prompts from our prompt catalog, and finally information aggregation using message passing from conventional GNN architectures. We explore several heuristics-based active node selection strategies to limit the computational and memory footprint of LLMs when handling millions of nodes. Through extensive experiments & ablation on popular graph benchmarks of varying scales (Cora, PubMed, ArXiv, & Products), we illustrate the effectiveness of our E-LLaGNN framework and reveal many interesting capabilities such as improved gradient flow in deep GNNs, LLM-free inference ability etc.

E-commerce platforms support the needs and livelihoods of their two most important stakeholders -- customers and producers/sellers. Multiple algorithmic systems, like ``search'' systems mediate the interactions between these stakeholders by connecting customers to producers with relevant items. Search results include (i) private label (PL) products that are manufactured/sold by the platform itself, as well as (ii) third-party products on advertised / sponsored and organic positions. In this paper, we systematically quantify the extent of PL product promotion on e-commerce search results for the two largest e-commerce platforms operating in India -- Amazon.in and Flipkart. By analyzing snapshots of search results across the two platforms, we discover high PL promotion on the initial result pages (~ 15% PLs are advertised on the first SERP of Amazon). Both platforms use different strategies to promote their PL products, such as placing more PLs on the advertised positions -- while Amazon places them on the first, middle, and last rows of the search results, Flipkart places them on the first two positions and the (entire) last column of the search results. We discover that these product placement strategies of both platforms conform with existing user attention strategies proposed in the literature. Finally, to supplement the findings from the collected data, we conduct a survey among 68 participants on Amazon Mechanical Turk. The click pattern from our survey shows that users strongly prefer to click on products placed at positions that correspond to the PL products on the search results of Amazon, but not so strongly on Flipkart. The click-through rate follows previously proposed theoretically grounded user attention distribution patterns in a two-dimensional layout.

Heatwaves pose significant health risks, particularly due to prolonged exposure to high summer temperatures. Vulnerable groups, especially pedestrians and cyclists on sun-exposed sidewalks, motivate the development of a route planning method that incorporates somatosensory temperature effects through shade ratio consideration. This paper is the first to introduce a pipeline that utilizes segmentation foundation models to extract shaded areas from high-resolution satellite images. These areas are then integrated into a multi-layered road map, enabling users to customize routes based on a balance between distance and shade exposure, thereby enhancing comfort and health during outdoor activities. Specifically, we construct a graph-based representation of the road map, where links indicate connectivity and are updated with shade ratio data for dynamic route planning. This system is already implemented online, with a video demonstration, and will be specifically adapted to assist travelers during the 2024 Olympic Games in Paris., Comment: Paper accepted to CIKM24 demo track

Modern Large Language Models (LLMs) are composed of matrices with billions of elements, making their storage and processing quite demanding in terms of computational resources and memory usage. Being significantly large, such matrices can often be expressed in low-rank format with potential to relax resource requirements. Unlike prior works which focus on developing novel matrix decomposition algorithms, in this work we first study the emergence of low-rank structures across matrices within different layers of LLMs and establish a consequential relationship between the gradient dynamics and emerging low-rank expressiveness of matrices. Our findings reveal that different layers exhibit varying levels of converged low-rank structure, necessitating a non-uniform rank reduction across them to minimize performance drop due to compression. In view of that, we present Weight Low-Rank Projection (WeLore) that unifies weight compression and memory-efficient fine-tuning as ONE, in a data-agnostic and one-shot way. WeLore capitalizes the heavy-tail distribution of singular values to identify a suitable rank reduction ratio for matrices within LLMs. Going beyond only as a compression technique, WeLore categorizes weight matrices into Low-rank Components (LRCs) and Non-Low-rank Components (N-LRCs) based on their ability to express themselves as low-rank. Our gradient perspective and extensive experiments illustrate that LRCs tend to have better finetuning capabilities and can closely mimic (sometimes outperform) the training loss trajectory and performance of full-finetuning with notable memory and compute footprint reduction. For example, finetuning a 50\% compressed LLaMa-2 7B model using only a fraction of parameters in LRCs (WeLore) can outperform its full finetuning with ~3x better throughput and ~0.6x GPU requirement. Our codes are available at \url{https://github.com/VITA-Group/welore}

Training Large Language Models (LLMs) is memory-intensive due to the large number of parameters and associated optimization states. GaLore, a recent method, reduces memory usage by projecting weight gradients into a low-rank subspace without compromising performance. However, GaLore relies on time-consuming Singular Value Decomposition (SVD) operations to identify the subspace, and the frequent subspace updates lead to significant training time overhead. Moreover, GaLore offers minimal improvements in accuracy and efficiency compared to LoRA in more accessible fine-tuning scenarios. To address these limitations, we introduce Q-Galore, a novel approach that substantially reduces memory usage by combining quantization and low-rank projection, surpassing the benefits of GaLore. Our method is based on two key observations: (i) the gradient subspace exhibits diverse properties, with some layers converging early in training while others are subject to frequent changes; (ii) the projection matrices are highly resilient to low-bit quantization. Leveraging these insights, Q-GaLore adaptively updates the gradient subspace based on its convergence statistics, achieving comparable performance while significantly reducing the number of SVD operations. We maintain the projection matrices in INT4 format and weights in INT8 format, incorporating stochastic rounding to capture accumulated gradient information. This approach enables a high-precision training trajectory using only low-precision weights. We demonstrate that Q-GaLore achieves highly competitive performance with exceptional memory efficiency. At pre-training, Q-GaLore facilitates training a LLaMA-7B model from scratch on a single NVIDIA RTX 4060 Ti with only 16 GB memory. At fine-tuning, it reduces memory consumption by up to 50% compared to LoRA and GaLore, while consistently outperforming QLoRA at the same memory cost.

Turing instability in complex networks have been shown in the literature to be dominated by the distribution of the nodal degrees. The conditions for Turing instability have been derived with an explicit dependence on the eigenvalues of the Laplacian, which in turn depends on the network topology. This study reveals that apart from average degree of the network, another global network measure - the nodal clustering - also plays a crucial role. Analytical and numerical results are presented to show the importance of clustering for several network topologies ranging from the $\mathbb{S}^1$ / $\mathbb{H}^2$ hyperbolic geometric networks that enable modelling the naturally occurring clustering in real world networks, as well as the random and scale free networks, which are obtained as limiting cases of the $\mathbb{S}^1$ / $\mathbb{H}^2$ model. Analysis of eigenvector localization properties in these networks are shown to reveal distinct signatures that enable identifying the so called Turing patterns even in complex networks.

Recent advances in adversarial robustness rely on an abundant set of training data, where using external or additional datasets has become a common setting. However, in real life, the training data is often kept private for security and privacy issues, while only the pretrained weight is available to the public. In such scenarios, existing methods that assume accessibility to the original data become inapplicable. Thus we investigate the pivotal problem of data-free adversarial robustness, where we try to achieve adversarial robustness without accessing any real data. Through a preliminary study, we highlight the severity of the problem by showing that robustness without the original dataset is difficult to achieve, even with similar domain datasets. To address this issue, we propose DataFreeShield, which tackles the problem from two perspectives: surrogate dataset generation and adversarial training using the generated data. Through extensive validation, we show that DataFreeShield outperforms baselines, demonstrating that the proposed method sets the first entirely data-free solution for the adversarial robustness problem., Comment: ICML 2024

Despite their remarkable successes, state-of-the-art large language models (LLMs), including vision-and-language models (VLMs) and unimodal language models (ULMs), fail to understand precise semantics. For example, semantically equivalent sentences expressed using different lexical compositions elicit diverging representations. The degree of this divergence and its impact on encoded semantics is not very well understood. In this paper, we introduce the SUGARCREPE++ dataset to analyze the sensitivity of VLMs and ULMs to lexical and semantic alterations. Each sample in SUGARCREPE++ dataset consists of an image and a corresponding triplet of captions: a pair of semantically equivalent but lexically different positive captions and one hard negative caption. This poses a 3-way semantic (in)equivalence problem to the language models. We comprehensively evaluate VLMs and ULMs that differ in architecture, pre-training objectives and datasets to benchmark the performance of SUGARCREPE++ dataset. Experimental results highlight the difficulties of VLMs in distinguishing between lexical and semantic variations, particularly in object attributes and spatial relations. Although VLMs with larger pre-training datasets, model sizes, and multiple pre-training objectives achieve better performance on SUGARCREPE++, there is a significant opportunity for improvement. We show that all the models which achieve better performance on compositionality datasets need not perform equally well on SUGARCREPE++, signifying that compositionality alone may not be sufficient for understanding semantic and lexical alterations. Given the importance of the property that the SUGARCREPE++ dataset targets, it serves as a new challenge to the vision-and-language community., Comment: Added the dataset link to the abstract

Air pollution kills 7 million people annually. The brick manufacturing industry accounts for 8%-14% of air pollution in the densely populated Indo-Gangetic plain. Due to the unorganized nature of brick kilns, policy violation detection, such as proximity to human habitats, remains challenging. While previous studies have utilized computer vision-based machine learning methods for brick kiln detection from satellite imagery, they utilize proprietary satellite data and rarely focus on compliance with government policies. In this research, we introduce a scalable framework for brick kiln detection and automatic compliance monitoring. We use Google Maps Static API to download the satellite imagery followed by the YOLOv8x model for detection. We identified and hand-verified 19579 new brick kilns across 9 states within the Indo-Gangetic plain. Furthermore, we automate and test the compliance to the policies affecting human habitats, rivers and hospitals. Our results show that a substantial number of brick kilns do not meet the compliance requirements. Our framework offers a valuable tool for governments worldwide to automate and enforce policy regulations for brick kilns, addressing critical environmental and public health concerns., Comment: The PI was not in favor of making the work public on arXiv as the content is not yet ready to be released

The phoretic Brownian dynamics method is shown here to be an effective approach to simulate the properties of colloidal chemophoretic based systems. The method is then optimized to allow for the comparison with results from multiparticle collision dynamics, a hydrodynamic method with explicit solvent, which can also be employed in the case of chemoattractive polymers. In order to obtain a good match of the conformational equilibrium properties of the models without and with explicit solvent, we propose a modified version of the phoretic Brownian dynamics accounting for the explicit solvent induced swelling. In the presence of activity, chemoattractive polymers show a transition to a compact globular state and hydrodynamics have a non-trivial influence in the polymer collapse times. The phoretic Brownian method can then be applied to much longer polymers, which allows the observation of a non-monotonous growth of both, the radius of gyration and the relaxation time with polymer length, for such chemoattractive active polymers., Comment: 10 pages, 8 figures

In this article, we will explore the fundamental concepts, including various basic concepts on $d$-complex manifolds, along with several differential operators and examine the relationships between them. A $d$-K\"ahler manifold is a $d$-complex manifold equipped with a metric that satisfies a specific condition. We prove the Hodge decomposition theorem on compact $d$-K\"ahler manifolds, which establishes a crucial relationship between certain de-Rham cohomology groups and Dolbeault cohomology groups on a compact $d$-K\"ahler manifold ., Comment: 39 pages, any comments are welcome