Search

Sanskrit, one of humanity's most ancient languages, has a vast collection of books and manuscripts on diverse topics that have been accumulated over millennia. However, its digital content (audio and text), which is vital for the training of AI systems, is profoundly limited. Furthermore, its intricate linguistics make it hard to develop robust NLP tools for wider accessibility. Given these constraints, we have developed an automatic speech recognition model for Sanskrit by employing transfer learning mechanism on OpenAI's Whisper model. After carefully optimising the hyper-parameters, we obtained promising results with our transfer-learned model achieving a word error rate of 15.42% on Vaksancayah dataset. An online demo of our model is made available for the use of public and to evaluate its performance firsthand thereby paving the way for improved accessibility and technological support for Sanskrit learning in the modern era., Comment: Paper has been accepted at the 4th International Conference on Computer, Communication, Control & Information Technology (C3IT), Hooghly, India, 2024, pp. 1-5

X-ray emissions due to the two-electron one-photon (TEOP) process in the neon projectile and aluminum target have been successfully observed for the beam energy window of 1.8-2.1 MeV. Experimental TEOP transition energies have been compared with theoretical predictions of flexible atomic structure code (FAC) and General-purpose Relativistic Atomic Structure (GRASP) package. Present results have been verified with reported theoretical and experimental values. Transition rates of the TEOP transitions have also been studied using the said codes. The observed lines have been assigned when the measured transition energies are in good agreement with the theoretical values. Such assignments have further been validated with the good agreements between the experimental and theoretical transition rates. Note that only the TEOP lines in projectile ions are seen with 1.8 MeV energy. In contrast, the TEOP lines in target ions are also observed well with 2.1 MeV energy. Thus, this study sheds useful light on the excitation mechanism of the TEOP processes in the low energy regimes., Comment: 6 pages, 2 figures, 3 tables. arXiv admin note: text overlap with arXiv:2201.02566

A significant challenge for computation offloading in wireless multi-hop networks is the complex interaction among traffic flows in the presence of interference. Existing approaches often ignore these key effects and/or rely on outdated queueing and channel state information. To fill these gaps, we reformulate joint offloading and routing as a routing problem on an extended graph with physical and virtual links. We adopt the state-of-the-art shortest path-biased Backpressure routing algorithm, which allows the destination and the route of a job to be dynamically adjusted at every time step based on network-wide long-term information and real-time states of local neighborhoods. In large networks, our approach achieves smaller makespan than existing approaches, such as separated Backpressure offloading and joint offloading and routing based on linear programming., Comment: 5 pages, 7 figures, accepted to IEEE ICASSP 2025

Graph neural networks (GNNs) are powerful tools for developing scalable, decentralized artificial intelligence in large-scale networked systems, such as wireless networks, power grids, and transportation networks. Currently, GNNs in networked systems mostly follow a paradigm of `centralized training, distributed execution', which limits their adaptability and slows down their development cycles. In this work, we fill this gap for the first time by developing a communication-efficient, fully distributed online training approach for GNNs applied to large networked systems. For a mini-batch with $B$ samples, our approach of training an $L$-layer GNN only adds $L$ rounds of message passing to the $LB$ rounds required by GNN inference, with doubled message sizes. Through numerical experiments in graph-based node regression, power allocation, and link scheduling in wireless networks, we demonstrate the effectiveness of our approach in training GNNs under supervised, unsupervised, and reinforcement learning paradigms.

Deep neural networks trained via empirical risk minimisation often exhibit significant performance disparities across groups, particularly when group and task labels are spuriously correlated (e.g., "grassy background" and "cows"). Existing bias mitigation methods that aim to address this issue often either rely on group labels for training or validation, or require an extensive hyperparameter search. Such data and computational requirements hinder the practical deployment of these methods, especially when datasets are too large to be group-annotated, computational resources are limited, and models are trained through already complex pipelines. In this paper, we propose Targeted Augmentations for Bias Mitigation (TAB), a simple hyperparameter-free framework that leverages the entire training history of a helper model to identify spurious samples, and generate a group-balanced training set from which a robust model can be trained. We show that TAB improves worst-group performance without any group information or model selection, outperforming existing methods while maintaining overall accuracy., Comment: Accepted at the 18th European Conference on Computer Vision (ECCV 2024) as an Oral presentation

A mixture of streaming and short-lived traffic presents a common yet challenging scenario for Backpressure routing in wireless multi-hop networks. Although state-of-the-art shortest-path biased backpressure (SP-BP) can significantly improve the latency of backpressure routing while retaining throughput optimality, it still suffers from the last-packet problem due to its inherent per-commodity queue structure and link capacity assignment. To address this challenge, we propose Ant Backpressure (Ant-BP), a fully distributed routing scheme that incorporates the multi-path routing capability of SP-BP into ant colony optimization (ACO) routing, which allows packets of different commodities to share link capacity in a first-in-first-out (FIFO) manner. Numerical evaluations show that Ant-BP can improve the latency and delivery ratio over SP-BP and ACO routing schemes, while achieving the same throughput of SP-BP under low-to-medium traffic loads.

In order to deploy deep neural networks (DNNs) in high-stakes scenarios, it is imperative that DNNs provide inference robust to external perturbations - both intentional and unintentional. Although the resilience of DNNs to intentional and unintentional perturbations has been widely investigated, a unified vision of these inherently intertwined problem domains is still missing. In this work, we fill this gap by providing a survey of the state of the art and highlighting the similarities of the proposed approaches.We also analyze the research challenges that need to be addressed to deploy resilient and secure DNNs. As there has not been any such survey connecting the resilience of DNNs to intentional and unintentional perturbations, we believe this work can help advance the frontier in both domains by enabling the exchange of ideas between the two communities.

Given a graph $G = (V, E)$ and an integer $k$, the Minimum Membership Dominating Set problem asks to compute a set $S \subseteq V$ such that for each $v \in V$, $1 \leq |N[v] \cap S| \leq k$. The problem is known to be NP-complete even on split graphs and planar bipartite graphs. In this paper, we approach the problem from the algorithmic standpoint and obtain several interesting results. We give an $\mathcal{O}^*(1.747^n)$ time algorithm for the problem on split graphs. Following a reduction from a special case of 1-in-3 SAT problem, we show that there is no sub-exponential time algorithm running in time $\mathcal{O}^*(2^{o(n)})$ for bipartite graphs, for any $k \geq 2$. We also prove that the problem is NP-complete when $\Delta = k+2$, for any $k\geq 5$, even for bipartite graphs. We investigate the parameterized complexity of the problem for the parameter twin cover and the combined parameter distance to cluster, membership($k$) and prove that the problem is fixed-parameter tractable. Using a dynamic programming based approach, we obtain a linear-time algorithm for trees., Comment: 23 pages, 6 figures

Automatic Sign Language (SL) recognition is an important task in the computer vision community. To build a robust SL recognition system, we need a considerable amount of data which is lacking particularly in Indian sign language (ISL). In this paper, we introduce a large-scale isolated ISL dataset and a novel SL recognition model based on skeleton graph structure. The dataset covers 2002 daily used common words in the deaf community recorded by 20 (10 male and 10 female) deaf adult signers (contains 40033 videos). We propose a SL recognition model namely Hierarchical Windowed Graph Attention Network (HWGAT) by utilizing the human upper body skeleton graph. The HWGAT tries to capture distinctive motions by giving attention to different body parts induced by the human skeleton graph. The utility of the proposed dataset and the usefulness of our model are evaluated through extensive experiments. We pre-trained the proposed model on the presented dataset and fine-tuned it across different sign language datasets further boosting the performance of 1.10, 0.46, 0.78, and 6.84 percentage points on INCLUDE, LSA64, AUTSL and WLASL respectively compared to the existing state-of-the-art keypoints-based models.

To reduce the latency of Backpressure (BP) routing in wireless multi-hop networks, we propose to enhance the existing shortest path-biased BP (SP-BP) and sojourn time-based backlog metrics, since they introduce no additional time step-wise signaling overhead to the basic BP. Rather than relying on hop-distance, we introduce a new edge-weighted shortest path bias built on the scheduling duty cycle of wireless links, which can be predicted by a graph convolutional neural network based on the topology and traffic of wireless networks. Additionally, we tackle three long-standing challenges associated with SP-BP: optimal bias scaling, efficient bias maintenance, and integration of delay awareness. Our proposed solutions inherit the throughput optimality of the basic BP, as well as its practical advantages of low complexity and fully distributed implementation. Our approaches rely on common link features and introduces only a one-time constant overhead to previous SP-BP schemes, or a one-time overhead linear in the network size to the basic BP. Numerical experiments show that our solutions can effectively address the major drawbacks of slow startup, random walk, and the last packet problem in basic BP, improving the end-to-end delay of existing low-overhead BP algorithms under various settings of network traffic, interference, and mobility., Comment: 16 pages, 15 figures, accepted for publication in IEEE Transactions on Machine Learning in Communications and Networking. arXiv admin note: text overlap with arXiv:2310.04364, arXiv:2211.10748

The most recent advances in medical imaging that have transformed diagnosis, especially in the case of interpreting X-ray images, are actively involved in the healthcare sector. The advent of digital image processing technology and the implementation of deep learning models such as Convolutional Neural Networks (CNNs) have made the analysis of X-rays much more accurate and efficient. In this article, some essential techniques such as edge detection, region-growing technique, and thresholding approach, and the deep learning models such as variants of YOLOv8-which is the best object detection and segmentation framework-are reviewed. We further investigate that the traditional image processing techniques like segmentation are very much simple and provides the alternative to the advanced methods as well. Our review gives useful knowledge on the practical usage of the innovative and traditional approaches of manual X-ray interpretation. The discovered information will help professionals and researchers to gain more profound knowledge in digital interpretation techniques in medical imaging., Comment: 29 pages, 20 figures

The information flows among the people while they communicate through social media websites. Due to the dependency on digital media, a person shares important information or regular updates with friends and family. The set of persons on social media forms a social network. Influence Maximization (IM) is a known problem in social networks. In social networks, information flows from one person to another using an underlying diffusion model. There are two fundamental diffusion models: the Independent Cascade Model (ICM) and the Linear Threshold Model (LTM). In this paper, we study a variant of the IM problem called Minimum Influential Seeds (MINFS) problem proposed by Qiang et al.[16]. It generalizes the classical IM problem with LTM as the diffusion model. Compared to IM, this variant has additional parameters: the influence threshold for each node and the propagation range. The propagation range is a positive integer that specifies how far the information can propagate from a node. A node on the network is not immediately influenced until it receives the same information from enough number of neighbors (influence threshold). Similarly, any node does not forward information until it receives the same information from a sufficient number of neighbors (activation threshold). Once a node becomes activated, it tries to activate or influence its neighbors. The MINFS problem aims to select the minimum number of initial spreader nodes such that all nodes of the graph are influenced. In this paper, we extend the study of the MINFS problem. We propose heuristics that construct seed sets based on the average degree of non-activated nodes, closest first, and backbone-based heaviest path.

Supervised learning methods have been found to exhibit inductive biases favoring simpler features. When such features are spuriously correlated with the label, this can result in suboptimal performance on minority subgroups. Despite the growing popularity of methods which learn from unlabeled data, the extent to which these representations rely on spurious features for prediction is unclear. In this work, we explore the impact of spurious features on Self-Supervised Learning (SSL) for visual representation learning. We first empirically show that commonly used augmentations in SSL can cause undesired invariances in the image space, and illustrate this with a simple example. We further show that classical approaches in combating spurious correlations, such as dataset re-sampling during SSL, do not consistently lead to invariant representations. Motivated by these findings, we propose LateTVG to remove spurious information from these representations during pre-training, by regularizing later layers of the encoder via pruning. We find that our method produces representations which outperform the baselines on several benchmarks, without the need for group or label information during SSL.

Nowadays, organizations use viral marketing strategies to promote their products through social networks. It is expensive to directly send the product promotional information to all the users in the network. In this context, Kempe et al. \cite{kempe2003maximizing} introduced the Influence Maximization (IM) problem, which identifies $k$ most influential nodes (spreader nodes), such that the maximum number of people in the network adopts the promotional message. Many variants of the IM problem have been studied in the literature, namely, Perfect Evangelising Set (PES), Perfect Awareness Problem (PAP), etc. In this work, we propose a maximization version of PAP called the \IM{} problem. Different people have different levels of interest in a particular product. This is modeled by assigning an interest value to each node in the network. Then, the problem is to select $k$ initial spreaders such that the sum of the interest values of the people (nodes) who become aware of the message is maximized. We study the \IM{} problem under two popular diffusion models: the Linear Threshold Model (LTM) and the Independent Cascade Model (ICM). We show that the \IM{} problem is NP-Hard under LTM. We give linear programming formulation for the problem under LTM. We propose four heuristic algorithms for the \IM{} problem: \LBE{} (\LB{}), Maximum Degree First Heuristic (\MD{}), \PBE{} (\PB{}), and Maximum Profit Based Greedy Heuristic (\MP{}). Extensive experimentation has been carried out on many real-world benchmark data sets for both diffusion models. The results show that among the proposed heuristics, \MP{} performs better in maximizing the interest value.

Scanning Thermal Microscopy (SThM) has become an important measurement tool for characterizing the thermal properties of materials at the nanometer scale. This technique requires a SThM probe that combines an Atomic Force Microscopy (AFM) probe and a very sensitive resistive thermometry; the thermometer being located at the apex of the probe tip allows the mapping of temperature or thermal properties of nanostructured materials with very high spatial resolution. The high interest of the SThM technique in the field of thermal nanoscience currently suffers from a low temperature sensitivity despite its high spatial resolution. To address this challenge, we developed a high vacuum-based AFM system hosting a highly sensitive niobium nitride (NbN) SThM probe to demonstrate its unique performance. As a proof of concept, we utilized this custom-built system to carry out thermal measurements using the 3$\omega$ method. By measuring the $V_{3\omega}$ voltage on the NbN resistive thermometer in vacuum conditions we were able to determine the SThM probe's thermal conductance and thermal time constant. The performance of the probe is demonstrated by doing thermal measurements in-contact with a sapphire sample., Comment: 17 pages, 13 figures

Human gait has been shown to provide crucial motion cues for various applications. Recognizing patterns in human gait has been widely adopted in various application areas such as security, virtual reality gaming, medical rehabilitation, and ailment identification. Furthermore, wearable inertial sensors have been widely used for not only recording gait but also to predict users' demography. Machine Learning techniques such as deep learning, combined with inertial sensor signals, have shown promising results in recognizing patterns in human gait and estimate users' demography. However, the black-box nature of such deep learning models hinders the researchers from uncovering the reasons behind the model's predictions. Therefore, we propose leveraging deep learning and Layer-Wise Relevance Propagation (LRP) to identify the important variables that play a vital role in identifying the users' demography such as age and gender. To assess the efficacy of this approach we train a deep neural network model on a large sensor-based gait dataset consisting of 745 subjects to identify users' age and gender. Using LRP we identify the variables relevant for characterizing the gait patterns. Thus, we enable interpretation of non-linear ML models which are experts in identifying the users' demography based on inertial signals. We believe this approach can not only provide clinicians information about the gait parameters relevant to age and gender but also can be expanded to analyze and diagnose gait disorders.

The key premise of federated learning (FL) is to train ML models across a diverse set of data-owners (clients), without exchanging local data. An overarching challenge to this date is client heterogeneity, which may arise not only from variations in data distribution, but also in data quality, as well as compute/communication latency. An integrated view of these diverse and concurrent sources of heterogeneity is critical; for instance, low-latency clients may have poor data quality, and vice versa. In this work, we propose FLASH(Federated Learning Across Simultaneous Heterogeneities), a lightweight and flexible client selection algorithm that outperforms state-of-the-art FL frameworks under extensive sources of heterogeneity, by trading-off the statistical information associated with the client's data quality, data distribution, and latency. FLASH is the first method, to our knowledge, for handling all these heterogeneities in a unified manner. To do so, FLASH models the learning dynamics through contextual multi-armed bandits (CMAB) and dynamically selects the most promising clients. Through extensive experiments, we demonstrate that FLASH achieves substantial and consistent improvements over state-of-the-art baselines -- as much as 10% in absolute accuracy -- thanks to its unified approach. Importantly, FLASH also outperforms federated aggregation methods that are designed to handle highly heterogeneous settings and even enjoys a performance boost when integrated with them.

We propose a learning-based framework for efficient power allocation in ad hoc interference networks under episodic constraints. The problem of optimal power allocation -- for maximizing a given network utility metric -- under instantaneous constraints has recently gained significant popularity. Several learnable algorithms have been proposed to obtain fast, effective, and near-optimal performance. However, a more realistic scenario arises when the utility metric has to be optimized for an entire episode under time-coupled constraints. In this case, the instantaneous power needs to be regulated so that the given utility can be optimized over an entire sequence of wireless network realizations while satisfying the constraint at all times. Solving each instance independently will be myopic as the long-term constraint cannot modulate such a solution. Instead, we frame this as a constrained and sequential decision-making problem, and employ an actor-critic algorithm to obtain the constraint-aware power allocation at each step. We present experimental analyses to illustrate the effectiveness of our method in terms of superior episodic network-utility performance and its efficiency in terms of time and computational complexity., Comment: ASILOMAR 2023

Parameter-efficient tuning (PET) methods such as LoRA, Adapter, and Visual Prompt Tuning (VPT) have found success in enabling adaptation to new domains by tuning small modules within a transformer model. However, the number of domains encountered during test time can be very large, and the data is usually unlabeled. Thus, adaptation to new domains is challenging; it is also impractical to generate customized tuned modules for each such domain. Toward addressing these challenges, this work introduces PLUTO: a Plug-and-pLay modUlar Test-time domain adaptatiOn strategy. We pre-train a large set of modules, each specialized for different source domains, effectively creating a ``module store''. Given a target domain with few-shot unlabeled data, we introduce an unsupervised test-time adaptation (TTA) method to (1) select a sparse subset of relevant modules from this store and (2) create a weighted combination of selected modules without tuning their weights. This plug-and-play nature enables us to harness multiple most-relevant source domains in a single inference call. Comprehensive evaluations demonstrate that PLUTO uniformly outperforms alternative TTA methods and that selecting $\leq$5 modules suffice to extract most of the benefit. At a high level, our method equips pre-trained transformers with the capability to dynamically adapt to new domains, motivating a new paradigm for efficient and scalable domain adaptation.

We propose a joint channel estimation and data detection algorithm for massive multilple-input multiple-output systems based on diffusion models. Our proposed method solves the blind inverse problem by sampling from the joint posterior distribution of the symbols and channels and computing an approximate maximum a posteriori estimation. To achieve this, we construct a diffusion process that models the joint distribution of the channels and symbols given noisy observations, and then run the reverse process to generate the samples. A unique contribution of the algorithm is to include the discrete prior distribution of the symbols and a learned prior for the channels. Indeed, this is key as it allows a more efficient exploration of the joint search space and, therefore, enhances the sampling process. Through numerical experiments, we demonstrate that our method yields a lower normalized mean squared error than competing approaches and reduces the pilot overhead.

We report a comprehensive study on the electronic transport properties of SrZn2Ge2 single crystals. The electrical resistivity of the compound exhibits metallic behavior, following a T^2 dependence below 35 K, consistent with the Fermi liquid behavior. However, a notable deviation is observed from this behavior at lower temperatures as a pronounced resistivity plateau emerges below 10 K. This plateau is remarkably robust, and persists under the magnetic fields of up to 10 T. Both the transverse and longitudinal magnetoresistance exhibit a crossover at critical field B* from weak-field quadratic-like to high-field unsaturated linear field dependence at low temperatures (T \leq 50 K). Possible sources of linear magnetoresistance are discussed based on the Fermi surface topology, classical and quantum transport models. The Hall resistivity data establish SrZn2Ge2 as a multiband system with contributions from both the electrons and holes. The Hall coefficient is observed to decrease with increasing temperature and magnetic field, changing its sign from positive to negative. The negative Hall coefficient observed at low temperatures in high fields and at high temperatures over the entire field range suggests that the highly mobile electron charge carriers dominate the electronic transport. Our first-principles calculations show that nontrivial topological surface states exist in SrZn2Ge2 within the bulk gap along the Gamma-M path. Notably, these surface states extend from the valence to conduction band with their number varying based on the Sr and Ge termination plane. The Fermi surface of the compound exhibits a distinct tetragonal petal-like structure, with one open and several closed surfaces. Overall, these findings offer crucial insights into the mechanisms underlying the electronic transport of the compound., Comment: 16 pages, 11 figures