Search

In this thesis, I evaluate the performance of Large Language Models (LLMs) on the Law School Admissions Test (LSAT), specifically the Logic Games section of the test. I focus on this section because it presents a complex logical reasoning task and thus is a valuable source of data for evaluating how modern, increasingly capable LLMs can handle hard logical reasoning tasks. I construct a dataset of LSAT logic games and their associated metadata, and extensively evaluate LLMs' performance in a Chain-of-Thought prompting setting. Given the weak performance in this setting, I explore other prompting frameworks on a smaller subset of the dataset, adapting ideas from Reflexion to this task. This results in a substantially improved accuracy of 70 percent for GPT-4 and 46 percent for GPT-3.5 on this data subset, highlighting the capacity of LLMs to revise their logical errors, despite initially weak performance. Finally, I analyze the types of logic games that models perform better or worse on, as well as the types of logical errors I observe from human annotation, providing detailed insights on the logical reasoning capabilities of LLMs., Comment: Bachelor's thesis. Dataset available on huggingface: https://huggingface.co/datasets/saumyamalik/lsat_logic_games-analytical_reasoning

Ensemble Kalman Filtering (EnKF) is a popular technique for data assimilation, with far ranging applications. However, the vanilla EnKF framework is not well-defined when perturbations are nonlinear. We study two non-linear extensions of the vanilla EnKF - dubbed the conditional-Gaussian EnKF (CG-EnKF) and the normal score EnKF (NS-EnKF) - which sidestep assumptions of linearity by constructing the Kalman gain matrix with the `conditional Gaussian' update formula in place of the traditional one. We then compare these models against a state-of-the-art deep learning based particle filter called the score filter (SF). This model uses an expensive score diffusion model for estimating densities and also requires a strong assumption on the perturbation operator for validity. In our comparison, we find that CG-EnKF and NS-EnKF dramatically outperform SF for a canonical problem in high-dimensional multiscale data assimilation given by the Lorenz-96 system. Our analysis also demonstrates that the CG-EnKF and NS-EnKF can handle highly non-Gaussian additive noise perturbations, with the latter typically outperforming the former.

This paper introduces a learning-based low-level controller for quadcopters, which adaptively controls quadcopters with significant variations in mass, size, and actuator capabilities. Our approach leverages a combination of imitation learning and reinforcement learning, creating a fast-adapting and general control framework for quadcopters that eliminates the need for precise model estimation or manual tuning. The controller estimates a latent representation of the vehicle's system parameters from sensor-action history, enabling it to adapt swiftly to diverse dynamics. Extensive evaluations in simulation demonstrate the controller's ability to generalize to unseen quadcopter parameters, with an adaptation range up to 16 times broader than the training set. In real-world tests, the controller is successfully deployed on quadcopters with mass differences of 3.7 times and propeller constants varying by more than 100 times, while also showing rapid adaptation to disturbances such as off-center payloads and motor failures. These results highlight the potential of our controller in extreme adaptation to simplify the design process and enhance the reliability of autonomous drone operations in unpredictable environments. The video and code are at: https://github.com/muellerlab/xadapt_ctrl, Comment: 12 pages, 9 figures

Foundation models, while highly effective, are often resource-intensive, requiring substantial inference time and memory. This paper addresses the challenge of making these models more accessible with limited computational resources by exploring dimensionality reduction techniques. Our goal is to enable users to run large pre-trained foundation models on standard GPUs without sacrificing performance. We investigate classical methods such as Principal Component Analysis alongside neural network-based adapters, aiming to reduce the dimensionality of multivariate time series data while preserving key features. Our experiments show up to a 10x speedup compared to the baseline model, without performance degradation, and enable up to 4.5x more datasets to fit on a single GPU, paving the way for more user-friendly and scalable foundation models., Comment: The first two authors contributed equally

We have derived precise analytic expressions for the quasinormal modes of test scalar, and Dirac fields in the background of the dilaton black hole. To achieve this, we employ the higher-order WKB expansion in terms of $1/\ell$. A comparison between the analytic formulas and time-domain integration reveals that the analytic approach generally yields more accurate results than the numerical results previously published using the lower-order WKB approach. We demonstrate that in the eikonal regime, test fields adhere to the correspondence between null geodesics and eikonal quasinormal modes., Comment: 5 figures, revtex

We present an approach to learn general robot manipulation priors from 3D hand-object interaction trajectories. We build a framework to use in-the-wild videos to generate sensorimotor robot trajectories. We do so by lifting both the human hand and the manipulated object in a shared 3D space and retargeting human motions to robot actions. Generative modeling on this data gives us a task-agnostic base policy. This policy captures a general yet flexible manipulation prior. We empirically demonstrate that finetuning this policy, with both reinforcement learning (RL) and behavior cloning (BC), enables sample-efficient adaptation to downstream tasks and simultaneously improves robustness and generalizability compared to prior approaches. Qualitative experiments are available at: \url{https://hgaurav2k.github.io/hop/}.

We perform the constraint analysis of a three (2 + 1)-dimensional (3D) field-theoretic example for Hodge theory $(i)$ at the classical level within the ambit of Lagrangian formulation, and $(ii)$ at the quantum level within the framework of Becchi-Rouet-Stora-Tyutin (BRST) formalism. We derive the conserved charges corresponding to the six continuous symmetries of our present theory. These six continuous summery transformations are the nilpotent (anti-)BRST and (anti-)co-BRST symmetries, a unique bosonic symmetry and the ghost-scale symmetry. It turns out that the Noether conserved (anti-)BRST charges are found to be non-nilpotent even though they are derived from the off-shell nilpotent versions of the continuous and infinitesimal (anti-)BRST symmetry transformations. We obtain the nilpotent versions of the (anti-)BRST charges from the non-nilpotent Noether (anti-)BRST charges and discuss the physicality criteria w.r.t. the latter to demonstrate that the operator forms of the first-class constraints (of the classical gauge theory) annihilate the physical states at the quantum level. This observation is consistent with Dirac's quantization conditions for the systems that are endowed with the constraints. We lay emphasis on the existence of a single (anti-)BRST invariant Curci-Ferrari (CF) type restriction in our theory and derive it from various theoretical angles., Comment: LaTeX file, 38 pages, version to appear in NPB

This paper presents, for the first time, an open-source implementation of the 3GPP Uplink Time Difference of Arrival (UL-TDoA) positioning method using the OpenAirInterface (OAI) framework. UL-TDoA is a critical positioning technique in 5G networks, leveraging the time differences of signal arrival at multiple base stations to determine the precise location of User Equipment (UE). This implementation aims to democratize access to advanced positioning technology by integrating UL-TDoA capabilities into both the Radio Access Network (RAN) and Core Network (CN) components of OAI, providing a comprehensive and 3GPP-compliant solution. The development includes the incorporation of essential protocol procedures, message flows, and interfaces as defined by 3GPP standards. Validation is conducted using two distinct methods: an OAI-RF simulator-based setup for controlled testing and an O-RAN-based Localization Testbed at EURECOM in real-world conditions. The results demonstrate the viability of this open-source UL-TDoA implementation, enabling precise positioning in various environments. By making this implementation publicly available, the study paves the way for widespread research, development, and innovation in the field of 5G positioning technologies, fostering collaboration and accelerating the advancement of cellular network positioning.

This paper presents a catalog of the energy distribution in the spectra of 263 stars in the wavelength range from 0.4 to 100 {\mu}m, which are at late stages of evolution and have been observed by the ISO space observatory. For each object in the catalog, estimates of the observed bolometric fluxes were derived from smoothed energy distribution curves. The catalog is available at https://infra.sai.msu.ru/sai\_lss\_sed both as a table and in machine-readable format. It is shown that for the specified sample of objects their ISO SWS spectra in the range 2.4-45 {\mu}m only 60% of cases correspond to the general shape of the continuum, and can be used without recalibration. A selection of carbon stars, accessible for the infrared observations from the MSU observatories has been made. For some of them the first brightness estimates in the K, L, and M bands were obtained with the new IR camera of the 2.5-m telescope of CMO., Comment: 12 pages, 3 figures

With the deeper penetration of inverter-based resources in power systems, false data injection attacks (FDIA) are a growing cyber-security concern. They have the potential to disrupt the system's stability like frequency stability, thereby leading to catastrophic failures. Therefore, an FDIA detection method would be valuable to protect power systems. FDIAs typically induce a discrepancy between the desired and the effective behavior of the power system dynamics. A suitable detection method can leverage power dynamics predictions to identify whether such a discrepancy was induced by an FDIA. This work investigates the efficacy of temporal and spatio-temporal state prediction models, such as Long Short-Term Memory (LSTM) and a combination of Graph Neural Networks (GNN) with LSTM, for predicting frequency dynamics in the absence of an FDIA but with noisy measurements, and thereby identify FDIA events. For demonstration purposes, the IEEE 39 New England Kron-reduced model simulated with a swing equation is considered. It is shown that the proposed state prediction models can be used as a building block for developing an effective FDIA detection method that can maintain high detection accuracy across various attack and deployment settings. It is also shown how the FDIA detection should be deployed to limit its exposure to detection inaccuracies and mitigate its computational burden., Comment: Under review

This paper asks to what extent social interaction influences one's behavior. We study this in the setting of two dancers dancing as a couple. We first consider a baseline in which we predict a dancer's future moves conditioned only on their past motion without regard to their partner. We then investigate the advantage of taking social information into account by conditioning also on the motion of their dancing partner. We focus our analysis on Swing, a dance genre with tight physical coupling for which we present an in-the-wild video dataset. We demonstrate that single-person future motion prediction in this context is challenging. Instead, we observe that prediction greatly benefits from considering the interaction partners' behavior, resulting in surprisingly compelling couple dance synthesis results (see supp. video). Our contributions are a demonstration of the advantages of socially conditioned future motion prediction and an in-the-wild, couple dance video dataset to enable future research in this direction. Video results are available on the project website: https://von31.github.io/synNsync

In recent years, financial analysts have been trying to develop models to predict the movement of a stock price index. The task becomes challenging in vague economic, social, and political situations like in Pakistan. In this study, we employed efficient models of machine learning such as long short-term memory (LSTM) and quantum long short-term memory (QLSTM) to predict the Karachi Stock Exchange (KSE) 100 index by taking monthly data of twenty-six economic, social, political, and administrative indicators from February 2004 to December 2020. The comparative results of LSTM and QLSTM predicted values of the KSE 100 index with the actual values suggested QLSTM a potential technique to predict stock market trends.

Recently, two models of quantum-corrected Schwarzschild-like black holes were developed within Effective Quantum Gravity, and the spectra of bosonic perturbations have been analyzed in several recent studies. In this work, we investigate the quasinormal modes of massless Dirac field perturbations around these black holes. By employing the higher-order WKB method and time-domain integration, we achieved consistency between the two approaches within their common range of validity. This concordance enables us to distinguish the two quantum-corrected black hole models from each other and from their Schwarzschild limit through their quasinormal spectra., Comment: 14 pages, 2 figures

The traditional viewpoint on Sparse Mixture of Experts (MoE) models is that instead of training a single large expert, which is computationally expensive, we can train many small experts. The hope is that if the total parameter count of the small experts equals that of the singular large expert, then we retain the representation power of the large expert while gaining computational tractability and promoting expert specialization. The recently introduced Soft MoE replaces the Sparse MoE's discrete routing mechanism with a differentiable gating function that smoothly mixes tokens. While this smooth gating function successfully mitigates the various training instabilities associated with Sparse MoE, it is unclear whether it induces implicit biases that affect Soft MoE's representation power or potential for expert specialization. We prove that Soft MoE with a single arbitrarily powerful expert cannot represent simple convex functions. This justifies that Soft MoE's success cannot be explained by the traditional viewpoint of many small experts collectively mimicking the representation power of a single large expert, and that multiple experts are actually necessary to achieve good representation power (even for a fixed total parameter count). Continuing along this line of investigation, we introduce a notion of expert specialization for Soft MoE, and while varying the number of experts yet fixing the total parameter count, we consider the following (computationally intractable) task. Given any input, how can we discover the expert subset that is specialized to predict this input's label? We empirically show that when there are many small experts, the architecture is implicitly biased in a fashion that allows us to efficiently approximate the specialized expert subset. Our method can be easily implemented to potentially reduce computation during inference., Comment: 21 pages, 5 figures, 13 tables

While inverter-based distributed energy resources (DERs) play a crucial role in integrating renewable energy into the power system, they concurrently diminish the grid's system inertia, elevating the risk of frequency instabilities. Furthermore, smart inverters, interfaced via communication networks, pose a potential vulnerability to cyber threats if not diligently managed. To proactively fortify the power grid against sophisticated cyber attacks, we propose to employ reinforcement learning (RL) to identify potential threats and system vulnerabilities. This study concentrates on analyzing adversarial strategies for false data injection, specifically targeting smart inverters involved in primary frequency control. Our findings demonstrate that an RL agent can adeptly discern optimal false data injection methods to manipulate inverter settings, potentially causing catastrophic consequences.

The use of high-dimensional systems for quantum communication opens interesting perspectives, such as increased information capacity and noise resilience. In this context, it is crucial to certify that a given quantum channel can reliably transmit high-dimensional quantum information. Here we develop efficient methods for the characterization of high-dimensional quantum channels. We first present a notion of dimensionality of quantum channels, and develop efficient certification methods for this quantity. We consider a simple prepare-and-measure setup, and provide witnesses for both a fully and a partially trusted scenario. In turn we apply these methods to a photonic experiment and certify dimensionalities up to 59 for a commercial graded-index multi-mode optical fiber. Moreover, we present extensive numerical simulations of the experiment, providing an accurate noise model for the fiber and exploring the potential of more sophisticated witnesses. Our work demonstrates the efficient characterization of high-dimensional quantum channels, a key ingredient for future quantum communication technologies.

Global environmental challenges and rising energy demands have led to extensive exploration of wind energy technologies. Accurate wind speed forecasting (WSF) is crucial for optimizing wind energy capture and ensuring system stability. However, predicting wind speed remains challenging due to its inherent randomness, fluctuation, and unpredictability. This study proposes the Temporal Convolutional Network Former (TCNFormer) for short-term (12-hour) wind speed forecasting. The TCNFormer integrates the Temporal Convolutional Network (TCN) and transformer encoder to capture the spatio-temporal features of wind speed. The transformer encoder consists of two distinct attention mechanisms: causal temporal multi-head self-attention (CT-MSA) and temporal external attention (TEA). CT-MSA ensures that the output of a step derives only from previous steps, i.e., causality. Locality is also introduced to improve efficiency. TEA explores potential relationships between different sample sequences in wind speed data. This study utilizes wind speed data from the NASA Prediction of Worldwide Energy Resources (NASA POWER) of Patenga Sea Beach, Chittagong, Bangladesh (latitude 22.2352{\deg} N, longitude 91.7914{\deg} E) over a year (six seasons). The findings indicate that the TCNFormer outperforms state-of-the-art models in prediction accuracy. The proposed TCNFormer presents a promising method for spatio-temporal WSF and may achieve desirable performance in real-world applications of wind power systems.

We propose self-similar contracting groups as a platform for cryptographic schemes based on simultaneous conjugacy search problem (SCSP). The class of these groups contains extraordinary examples like Grigorchuk group, which is known to be non-linear, thus making some of existing attacks against SCSP inapplicable. The groups in this class admit a natural normal form based on the notion of a nucleus portrait, that plays a key role in our approach. While for some groups in the class the conjugacy search problem has been studied, there are many groups for which no algorithms solving it are known. Moreover, there are some self-similar groups with undecidable conjugacy problem. We discuss benefits and drawbacks of using these groups in group-based cryptography and provide computational analysis of variants of the length-based attack on SCSP for some groups in the class, including Grigorchuk group, Basilica group, and others., Comment: 37 pages, 13 figures

Voice-cloning (VC) systems have seen an exceptional increase in the realism of synthesized speech in recent years. The high quality of synthesized speech and the availability of low-cost VC services have given rise to many potential abuses of this technology. Several detection methodologies have been proposed over the years that can detect voice spoofs with reasonably good accuracy. However, these methodologies are mostly evaluated on clean audio databases, such as ASVSpoof 2019. This paper evaluates SOTA Audio Spoof Detection approaches in the presence of laundering attacks. In that regard, a new laundering attack database, called the ASVSpoof Laundering Database, is created. This database is based on the ASVSpoof 2019 (LA) eval database comprising a total of 1388.22 hours of audio recordings. Seven SOTA audio spoof detection approaches are evaluated on this laundered database. The results indicate that SOTA systems perform poorly in the presence of aggressive laundering attacks, especially reverberation and additive noise attacks. This suggests the need for robust audio spoof detection., Comment: Conference Paper

Recent data from elliptical galaxies indicate that the growth in the masses of black holes exceeds what is expected from the accretion of surrounding matter, and this growth appears to be dependent on the expansion of the universe. This phenomenon can be explained by considering the accretion of dark energy, which is responsible for cosmological expansion, into these black holes. In this paper, we investigate the perturbative interaction of a black hole with a real scalar field $\phi$ (which can represent dark energy) in a cosmological background using an appropriate metric. We derive solutions for the field $\phi(t, r)$, the black hole mass $M(t)$, and the expansion rate $H(t)$, and discuss the behaviour of the scalar field $\phi$ in the vicinity of the black hole, with respect to exterior and interior observers. We obtain the energy density of $\phi$ inside of the black hole and find that energy density converges and tends to take a non-vanishing fixed values (stable) at late time of cosmological expansion. We find that on horizon surfaces, it is possible to make the field $\phi$ continuous and bounded (maybe differentiable) function., Comment: accepted in European Physical Journal C

This paper presents the External Attention Vision Transformer (EAViT) model, a novel approach designed to enhance audio classification accuracy. As digital audio resources proliferate, the demand for precise and efficient audio classification systems has intensified, driven by the need for improved recommendation systems and user personalization in various applications, including music streaming platforms and environmental sound recognition. Accurate audio classification is crucial for organizing vast audio libraries into coherent categories, enabling users to find and interact with their preferred audio content more effectively. In this study, we utilize the GTZAN dataset, which comprises 1,000 music excerpts spanning ten diverse genres. Each 30-second audio clip is segmented into 3-second excerpts to enhance dataset robustness and mitigate overfitting risks, allowing for more granular feature analysis. The EAViT model integrates multi-head external attention (MEA) mechanisms into the Vision Transformer (ViT) framework, effectively capturing long-range dependencies and potential correlations between samples. This external attention (EA) mechanism employs learnable memory units that enhance the network's capacity to process complex audio features efficiently. The study demonstrates that EAViT achieves a remarkable overall accuracy of 93.99%, surpassing state-of-the-art models.

We consider the problem of few-viewpoint 3D surface reconstruction using raw measurements from a lidar system. Lidar captures 3D scene geometry by emitting pulses of light to a target and recording the speed-of-light time delay of the reflected light. However, conventional lidar systems do not output the raw, captured waveforms of backscattered light; instead, they pre-process these data into a 3D point cloud. Since this procedure typically does not accurately model the noise statistics of the system, exploit spatial priors, or incorporate information about downstream tasks, it ultimately discards useful information that is encoded in raw measurements of backscattered light. Here, we propose to leverage raw measurements captured with a single-photon lidar system from multiple viewpoints to optimize a neural surface representation of a scene. The measurements consist of time-resolved photon count histograms, or transients, which capture information about backscattered light at picosecond time scales. Additionally, we develop new regularization strategies that improve robustness to photon noise, enabling accurate surface reconstruction with as few as 10 photons per pixel. Our method outperforms other techniques for few-viewpoint 3D reconstruction based on depth maps, point clouds, or conventional lidar as demonstrated in simulation and with captured data., Comment: WACV 2025. Project Page: https://weihan1.github.io/transientangelo/

Despite the recent popularity of Large Language Models (LLMs) in Machine Translation (MT), their performance in low-resource languages (LRLs) still lags significantly behind Neural Machine Translation (NMT) models. In this work, we explore what it would take to adapt LLMs for the low-resource setting. Particularly, we re-examine the role of two factors: a) the importance and application of parallel data, and b) diversity in Supervised Fine-Tuning (SFT). Recently, parallel data has seen reduced use in adapting LLMs for MT, while data diversity has been embraced to promote transfer across languages and tasks. However, for low-resource LLM-MT, we show that the opposite is true for both considerations: a) parallel data is critical during both pre-training and SFT; b) diversity tends to cause interference instead of transfer. Our experiments with three LLMs across two low-resourced language groups -- Indigenous American and North-East Indian -- reveal consistent trends, underscoring the generalizability of our findings. We believe these insights will be valuable for scaling to massively multilingual LLM-MT models that can effectively serve LRLs., Comment: 10 pages, 6 figures

Requirements Engineering (RE) plays a pivotal role in software development, encompassing tasks such as requirements elicitation, analysis, specification, and change management. Despite its critical importance, RE faces challenges including communication complexities, early-stage uncertainties, and accurate resource estimation. This study empirically investigates the effectiveness of utilizing Large Language Models (LLMs) to automate requirements analysis tasks. We implemented a multi-agent system that deploys AI models as agents to generate user stories from initial requirements, assess and improve their quality, and prioritize them using a selected technique. In our implementation, we deployed four models, namely GPT-3.5, GPT-4 Omni, LLaMA3-70, and Mixtral-8B, and conducted experiments to analyze requirements on four real-world projects. We evaluated the results by analyzing the semantic similarity and API performance of different models, as well as their effectiveness and efficiency in requirements analysis, gathering users' feedback on their experiences. Preliminary results indicate notable variations in task completion among the models. Mixtral-8B provided the quickest responses, while GPT-3.5 performed exceptionally well when processing complex user stories with a higher similarity score, demonstrating its capability in deriving accurate user stories from project descriptions. Feedback and suggestions from the four project members further corroborate the effectiveness of LLMs in improving and streamlining RE phases.

Magnetic nanotubes have garnered immense attention for their potential in high-density magnetic memory, owing to their stable flux closure configuration and fast, reproducible reversal processes. However, characterizing their magnetic configuration through straightforward methodologies remains a challenge in both scope and detail. Here, we elucidate the magnetic state details using Remanence Field Ferromagnetic Resonance Spectroscopy (RFMR) for arrays of electrodeposited nanotubes. Micromagnetic simulations revealed distinct spin configurations while coming from saturation, including the edge vortex, onion, uniform and curling states, with chirality variations depending on the preparation field direction. Dynamic measurements, coupled with RFMR spectra analysis, unveiled multiple FMR modes corresponding to these spin configurations. The evolution of spin configurations under bias fields were studied, indicating nucleation within the curling state. Observations revealed opposite RFMR spectra, denoting opposite magnetic spin configurations after removing the positive and negative saturating fields when the magnetic field was applied along {theta_H=0} and perpendicular {theta_H= 90} to the nanotube axis. We observed a mixture of the non-uniform curling states with the end vortex state (onion-like curling state) at the end of the nanotubes for the theta_H=0(90) and uniform magnetization states in the middle of the nanotubes for the theta_H=0 configuration. Building on RFMR information, frequency-swept FMR absorption spectra obtained at different bias fields allowed the characterization of magnetization states. This picture was supported by micromagnetic simulations. These findings were further substantiated with First Order Reversal Curve measurements (FORC).

Interpretable classification of time series presents significant challenges in high dimensions. Traditional feature selection methods in the frequency domain often assume sparsity in spectral density matrices (SDMs) or their inverses, which can be restrictive for real-world applications. In this article, we propose a model-based approach for classifying high-dimensional stationary time series by assuming sparsity in the difference between inverse SDMs. Our approach emphasizes the interpretability of model parameters, making it especially suitable for fields like neuroscience, where understanding differences in brain network connectivity across various states is crucial. The estimators for model parameters demonstrate consistency under appropriate conditions. We further propose using standard deep learning optimizers for parameter estimation, employing techniques such as mini-batching and learning rate scheduling. Additionally, we introduce a method to screen the most discriminatory frequencies for classification, which exhibits the sure screening property under general conditions. The flexibility of the proposed model allows the significance of covariates to vary across frequencies, enabling nuanced inferences and deeper insights into the underlying problem. The novelty of our method lies in the interpretability of the model parameters, addressing critical needs in neuroscience. The proposed approaches have been evaluated on simulated examples and the `Alert-vs-Drowsy' EEG dataset.

We introduce Imagen 3, a latent diffusion model that generates high quality images from text prompts. We describe our quality and responsibility evaluations. Imagen 3 is preferred over other state-of-the-art (SOTA) models at the time of evaluation. In addition, we discuss issues around safety and representation, as well as methods we used to minimize the potential harm of our models.

The groundbreaking performance of transformers in Natural Language Processing (NLP) tasks has led to their replacement of traditional Convolutional Neural Networks (CNNs), owing to the efficiency and accuracy achieved through the self-attention mechanism. This success has inspired researchers to explore the use of transformers in computer vision tasks to attain enhanced long-term semantic awareness. Vision transformers (ViTs) have excelled in various computer vision tasks due to their superior ability to capture long-distance dependencies using the self-attention mechanism. Contemporary ViTs like Data Efficient Transformers (DeiT) can effectively learn both global semantic information and local texture information from images, achieving performance comparable to traditional CNNs. However, their impressive performance comes with a high computational cost due to very large number of parameters, hindering their deployment on devices with limited resources like smartphones, cameras, drones etc. Additionally, ViTs require a large amount of data for training to achieve performance comparable to benchmark CNN models. Therefore, we identified two key challenges in deploying ViTs on smaller form factor devices: the high computational requirements of large models and the need for extensive training data. As a solution to these challenges, we propose compressing large ViT models using Knowledge Distillation (KD), which is implemented data-free to circumvent limitations related to data availability. Additionally, we conducted experiments on object detection within the same environment in addition to classification tasks. Based on our analysis, we found that datafree knowledge distillation is an effective method to overcome both issues, enabling the deployment of ViTs on less resourceconstrained devices.

We explore the infinite-horizon Distributionally Robust (DR) linear-quadratic control. While the probability distribution of disturbances is unknown and potentially correlated over time, it is confined within a Wasserstein-2 ball of a radius $r$ around a known nominal distribution. Our goal is to devise a control policy that minimizes the worst-case expected Linear-Quadratic Regulator (LQR) cost among all probability distributions of disturbances lying in the Wasserstein ambiguity set. We obtain the optimality conditions for the optimal DR controller and show that it is non-rational. Despite lacking a finite-order state-space representation, we introduce a computationally tractable fixed-point iteration algorithm. Our proposed method computes the optimal controller in the frequency domain to any desired fidelity. Moreover, for any given finite order, we use a convex numerical method to compute the best rational approximation (in $H_\infty$-norm) to the optimal non-rational DR controller. This enables efficient time-domain implementation by finite-order state-space controllers and addresses the computational hurdles associated with the finite-horizon approaches to DR-LQR problems, which typically necessitate solving a Semi-Definite Program (SDP) with a dimension scaling with the time horizon. We provide numerical simulations to showcase the effectiveness of our approach., Comment: Accepted at CDC 2024

In this paper we introduce and study the notion of I-convergence of sequences in a metric-like space, where I is an ideal of subsets of the set N of all natural numbers. Further introducing the notion of I*-convergence of sequences in a metric-like space we study its relationship with I-convergence.

The equation of state (EOS) for neutron stars is modeled using the Relativistic Mean Field (RMF) approach with a mesonic nonlinear (NL) interaction, a modified sigma cut potential (NL-$\sigma$ cut), and the influences of dark matter in the NL (NL DM). Using a Bayesian analysis framework, we evaluate the plausibility and impact of each scenario. Experimental constraints on the general properties of finite nuclei and heavy ion collisions, along with astrophysical observational data on neutron star radii and tidal deformation, have been taken into account. It was shown that all models, including the PREX-II data, were less favored, indicating that this experimental data seemed to be in tension with the other constraints included in the inference procedure, and were incompatible with chiral effective field theoretical calculations of pure neutron matter. Considering the models with no PREX-II constraints, we find the model NL-$\sigma$ cut with the largest Bayes evidence, indicating that the constraints considered favor the stiffening of the EOS at large densities. Conversely, the neutron star with a dark matter component is the least favorable case in light of recent observational constraints, among different scenarios considered here. The $f$ and $p$ modes were calculated within the Cowling approximation, and it can be seen that $f$ modes are sensitive to the EOS. An analysis of the slopes of the mass-radius curves and $f$-mode mass curves has indicated that these quantities may help distinguish the different scenarios.We also analyzed the impact of new PSR J0437-4715 measurements on neutron star mass-radius estimates, noting a $\sim$ 0.2 km reduction in the 90\% CI upper boundary across all models and a significant Bayes evidence decrease, indicating potential conflicts with previous data or the necessity for more adaptable models., Comment: 18 pages, 10 figures and 6 tables