arXiv subjects

Sourav Sahoo

Publications and source records attributed to Sourav Sahoo.

14 recordsLinked to original sources

arXiv2026

Learning to Bid in Discriminatory Auctions with Budget Constraints

We study repeated bidding in multi-unit discriminatory (pay-as-bid) auctions for a single bidder with per-round utility equal to value minus $\alpha$ times payment, where $\alpha\in[0,1]$ is a cost-of-capital parameter. The bidder aims to maximize cumulative utility over $T$ rounds subject to a total budget $B$. The problem is challenging even without budgets: the action space is exponential in $M$, the maximum demand of the bidder and the valuation vector (context) varies over time. Exploiting a decomposition of utility across units, we develop polynomial-time learning algorithms based on shortest paths in a directed acyclic graph, obtaining sublinear regret under both full-information and bandit feedback. In the bandit setting, the regret is independent of the number of contexts due to complete cross-learning: observing the utility of the chosen action under the realized context reveals the utility for the same action under all counterfactual contexts. With budget constraints, when the average normalized per-round budget $\rho=\frac{B}{MT}<1$, we design a coupled primal-dual algorithm in which the DAG-based procedure uses dual-adjusted edge weights for primal updates, while online gradient descent updates the dual variable, yielding $\rho$-approximate sublinear regret. Finally, we give implementations whose per-round time and space are independent of the number of contexts, enabling scalability to large or even infinite context spaces.

cs.LG↗

arXiv2025

Learning Safe Strategies for Value Maximizing Buyers in Uniform Price Auctions

We study the bidding problem in repeated uniform price multi-unit auctions from the perspective of a value-maximizing buyer. The buyer aims to maximize their cumulative value over $T$ rounds while adhering to per-round return-on-investment (RoI) constraints in a strategic (or adversarial) environment. Using an $m$-uniform bidding format, the buyer submits $m$ bid-quantity pairs $(b_i, q_i)$ to demand $q_i$ units at bid $b_i$, with $m \ll M$ in practice, where $M$ denotes the maximum demand of the buyer. We introduce the notion of safe bidding strategies as those that satisfy the RoI constraints irrespective of competing bids. Despite the stringent requirement, we show that these strategies satisfy a mild no-overbidding condition, depend only on the valuation curve of the bidder, and the bidder can focus on a finite subset without loss of generality. Though the subset size is $O(M^m)$, we design a polynomial-time learning algorithm that achieves sublinear regret, both in full-information and bandit settings, relative to the hindsight-optimal safe strategy. We assess the robustness of safe strategies against the hindsight-optimal strategy from a richer class. We define the richness ratio $α\in (0,1]$ as the minimum ratio of the value of the optimal safe strategy to that of the optimal strategy from richer class and construct hard instances showing the tightness of $α$. Our algorithm achieves $α$-approximate sublinear regret against these stronger benchmarks. Simulations on semi-synthetic auction data show that empirical richness ratios significantly outperform the theoretical worst-case bounds. The proposed safe strategies and learning algorithm extend naturally to more nuanced buyer and competitor models.

cs.DS↗

arXiv2023

Online Subset Selection using $α$-Core with no Augmented Regret

We revisit the classic problem of optimal subset selection in the online learning set-up. Assume that the set $[N]$ consists of $N$ distinct elements. On the $t$th round, an adversary chooses a monotone reward function $f_t: 2^{[N]} \to \mathbb{R}_+$ that assigns a non-negative reward to each subset of $[N].$ An online policy selects (perhaps randomly) a subset $S_t \subseteq [N]$ consisting of $k$ elements before the reward function $f_t$ for the $t$th round is revealed to the learner. As a consequence of its choice, the policy receives a reward of $f_t(S_t)$ on the $t$th round. Our goal is to design an online sequential subset selection policy to maximize the expected cumulative reward accumulated over a time horizon. In this connection, we propose an online learning policy called SCore (Subset Selection with Core) that solves the problem for a large class of reward functions. The proposed SCore policy is based on a new polyhedral characterization of the reward functions called $α$-Core - a generalization of Core from the cooperative game theory literature. We establish a learning guarantee for the SCore policy in terms of a new performance metric called $α$-augmented regret. In this new metric, the performance of the online policy is compared with an unrestricted offline benchmark that can select all $N$ elements at every round. We show that a large class of reward functions, including submodular, can be efficiently optimized with the SCore policy. We also extend the proposed policy to the optimistic learning set-up where the learner has access to additional untrusted hints regarding the reward functions. Finally, we conclude the paper with a list of open problems.

cs.LG↗

arXiv2022

Distributed Online Optimization with Byzantine Adversarial Agents

We study the problem of non-constrained, discrete-time, online distributed optimization in a multi-agent system where some of the agents do not follow the prescribed update rule either due to failures or malicious intentions. None of the agents have prior information about the identities of the faulty agents and any agent can communicate only with its immediate neighbours. At each time step, a locally Lipschitz strongly convex cost function is revealed locally to all the agents and the non-faulty agents update their states using their local information and the information obtained from their neighbours. We measure the performance of the online algorithm by comparing it to its offline version, when the cost functions are known apriori. The difference between the same is termed as regret. Under sufficient conditions on the graph topology, the number and location of the adversaries, the defined regret grows sublinearly. We further conduct numerical experiments to validate our theoretical results.

math.OC↗

arXiv2022

$k\texttt{-experts}$ -- Online Policies and Fundamental Limits

We introduce the $\texttt{$k$-experts}$ problem - a generalization of the classic Prediction with Expert's Advice framework. Unlike the classic version, where the learner selects exactly one expert from a pool of $N$ experts at each round, in this problem, the learner can select a subset of $k$ experts at each round $(1\leq k\leq N)$. The reward obtained by the learner at each round is assumed to be a function of the $k$ selected experts. The primary objective is to design an online learning policy with a small regret. In this pursuit, we propose $\texttt{SAGE}$ ($\textbf{Sa}$mpled Hed$\textbf{ge}$) - a framework for designing efficient online learning policies by leveraging statistical sampling techniques. For a wide class of reward functions, we show that $\texttt{SAGE}$ either achieves the first sublinear regret guarantee or improves upon the existing ones. Furthermore, going beyond the notion of regret, we fully characterize the mistake bounds achievable by online learning policies for stable loss functions. We conclude the paper by establishing a tight regret lower bound for a variant of the $\texttt{$k$-experts}$ problem and carrying out experiments with standard datasets.

cs.IT↗

arXiv2021

Multi-Modal Detection of Alzheimer's Disease from Speech and Text

Reliable detection of the prodromal stages of Alzheimer's disease (AD) remains difficult even today because, unlike other neurocognitive impairments, there is no definitive diagnosis of AD in vivo. In this context, existing research has shown that patients often develop language impairment even in mild AD conditions. We propose a multimodal deep learning method that utilizes speech and the corresponding transcript simultaneously to detect AD. For audio signals, the proposed audio-based network, a convolutional neural network (CNN) based model, predicts the diagnosis for multiple speech segments, which are combined for the final prediction. Similarly, we use contextual embedding extracted from BERT concatenated with a CNN-generated embedding for classifying the transcript. The individual predictions of the two models are then combined to make the final classification. We also perform experiments to analyze the model performance when Automated Speech Recognition (ASR) system generated transcripts are used instead of manual transcription in the text-based model. The proposed method achieves 85.3% 10-fold cross-validation accuracy when trained and evaluated on the Dementiabank Pitt corpus.

cs.LG↗

arXiv2021

Observation of Coherent Spin Waves in a Three-Dimensional Artificial Spin Ice Structure

Harnessing high-frequency spin dynamics in three-dimensional (3D) nanostructures may lead to paradigm-shifting, next generation devices including high density spintronics and neuromorphic systems. Despite remarkable progress in fabrication, the measurement and interpretation of spin dynamics in complex 3D structures remain exceptionally challenging. Here we take a first step and measure coherent spin waves within a 3D artificial spin ice (ASI) structure using Brillouin light scattering. The 3D-ASI was fabricated by using a combination of two-photon lithography and thermal evaporation. Two spin-wave modes were observed in the experiment whose frequencies showed a monotonic variation with the applied field strength. Numerical simulations qualitatively reproduced the observed modes. The simulated mode profiles revealed the collective nature of the modes extending throughout the complex network of nanowires while showing spatial quantization with varying mode quantization numbers. The study shows a well-defined means to explore high-frequency spin dynamics in complex 3D spintronic and magnonic structures.

cond-mat.mes-hall↗

arXiv2021

Looking Through Glass: Knowledge Discovery from Materials Science Literature using Natural Language Processing

Most of the knowledge in materials science literature is in the form of unstructured data such as text and images. Here, we present a framework employing natural language processing, which automates text and image comprehension and precision knowledge extraction from inorganic glasses' literature. The abstracts are automatically categorized using latent Dirichlet allocation (LDA), providing a way to classify and search semantically linked publications. Similarly, a comprehensive summary of images and plots are presented using the 'Caption Cluster Plot' (CCP), which provides direct access to the images buried in the papers. Finally, we combine the LDA and CCP with the chemical elements occurring in the manuscript to present an 'Elemental map', a topical and image-wise distribution of chemical elements in the literature. Overall, the framework presented here can be a generic and powerful tool to extract and disseminate material-specific information on composition-structure-processing-property dataspaces, allowing insights into fundamental problems relevant to the materials science community and accelerated materials discovery.

cs.DL↗

arXiv2020

Magnetization Dynamics of Nanoscale Magnetic Materials: A Perspective

Nanomagnets form the building blocks for a gamut of miniaturized energy-efficient devices including data storage, memory, wave-based computing, sensors and biomedical devices. They also offer a span of exotic phenomena and stern challenges. The progress in the magnetization dynamics of single nanomagnets and one- and two-dimensional arrays of nanostructures in the form of dots, antidots, nanoparticles, binary and bicomponent structures and patterned multilayers have been presented. Progress in unconventional and new structures like artificial spin ice and three-dimensional nanomagnets and spin textures like domain walls, vortex and skyrmions have been presented. Furthermore, a huge variety of new topics in the magnetization dynamics of magnetic nanostructures are rapidly emerging. An overview of the steadily evolving topics like spatio-temporal imaging of fast dynamics of nanostructures, dynamics of spin textures, artificial spin ice have been discussed. In addition, dynamics of contemporary and newly transpired magnetic architectures such as nanomagnet arrays with complex basis and symmetry, magnonic quasicrystals, fractals, defect structures, novel three-dimensional structures have been introduced. Effects of various spin-orbit coupling and ensuing spin textures as well as quantum hybrid systems comprising of magnon-photon, magnon-phonon and magnon-magnon coupling, antiferromagnetic nanostructures are rapidly growing and are expected to dominate this research field in the coming years. Finally, associated topics like nutation dynamics and nanomagnet antenna are briefly discussed. Despite showing a great progress, only a small fraction of nanomagnetism and its ancillary topics have been explored so far and huge efforts are envisaged in this evergrowing research area in the generations to come.

cond-mat.mes-hall↗

arXiv2020

Nanochannels for Spin-Wave Manipulation in Ni80Fe20 Nanodot Arrays

Patterned magnetic nanostructures are potential candidates for future energy efficient, on-chip communication devices. Here, we have experimentally and numerically studied the role of nanochannels to manipulate spin waves in Ni80Fe20 connected nanodot arrays of varying filling fraction. Rich spin-wave spectra are observed in these samples, where the number of spin-wave modes decreases with increasing filling fraction due to the retrenchment of the demagnetizing field. The nanochannels affect the spin-wave modes of the connected dots through dipole-exchange coupling. For all modes the vertical nanochannels couple the nanodots, except for the highest frequency modes where all nanochannels act as coupler. This feature is further explored in the simulation, which reveals that only the highest frequency mode can propagate through all the nanochannels, analogues to an electronic demultiplexer. This study will be useful to understand the role of nanochannels in patterned magnetic nanostructures and their applications in spin-wave based communication devices.

cond-mat.mes-hall↗

arXiv2020

The effect of material defects on resonant spin wave modes in a nanomagnet

We have theoretically studied how resonant spin wave modes in an elliptical nanomagnet are affected by fabrication defects, such as small local thickness variations. Our results indicate that defects of this nature, which can easily result from the fabrication process, or are sometimes deliberately introduced during the fabrication process, will significantly alter the frequencies, magnetic field dependence of the frequencies, and the power and phase profiles of the resonant spin wave modes. They can also spawn new resonant modes and quench existing ones. All this has important ramifications for multi-device circuits based on spin waves, such as phase locked oscillators for neuromorphic computing, where the device-to-device variability caused by defects can be inhibitory.

cond-mat.mes-hall↗

arXiv2019

Error Probability in Magneto-elastic Switching of Non-ideal Nanomagnets with Defects: A Case Study for the Viability of Straintronic Logic and Memory

Magneto-elastic (straintronic) switching of bistable magnetostrictive nanomagnets is an extremely energy-efficient switching methodology for (magnetic) binary switches that has recently attracted widespread attention because of its potential application in ultra-low-power digital computing hardware. Unfortunately, this modality of switching is also error very prone at room temperature. Theoretical studies of switching error probability of magneto-elastic switches have predicted probabilities ranging from 10E-8-10E-3 at room temperature for ideal, defect-free nanomagnets, but experiments with real nanomagnets show a much higher probability that exceeds 0.1 in some cases. The obvious spoilers that can cause this large difference are defects and non-idealities. Here, we have theoretically studied the effect of common defects (that occur during fabrication) on magneto-elastic switching probability in the presence of room-temperature thermal noise. Surprisingly, we found that even small defects increase the switching error probabilities by orders of magnitude. There is usually a critical stress that leads to the lowest error probability and its value increases enormously in the presence of defects. All this could limit or preclude the application of magneto-elastic (straintronic) binary switches in either Boolean logic or memory, despite their excellent energy-efficiency, and restrict them to non-Boolean (e.g. neuromorphic, stochastic) computing applications. We also studied the difference between magneto-elastic switching with a stress pulse of constant amplitude and sinusoidal time-varying amplitude (e.g. due to a surface acoustic wave) and found that the latter method is more reliable and generates lower switching error probabilities in most cases, provided the time variation is reasonably slow.

cond-mat.mes-hall↗

arXiv2019

Tunability of domain structure and magnonic spectra in antidot arrays of Heusler alloy

Materials suitable for magnonic crystals demand low magnetic damping and long spin wave (SW) propagation distance. In this context Co based Heusler compounds are ideal candidates for magnonic based applications. In this work, antidot arrays (with different shapes) of epitaxial $\mathrm{Co}_2\mathrm{Fe}_{0.4}\mathrm{Mn}_{0.6}\mathrm{Si}$ (CFMS) Heusler alloy thin films have been prepared using e-beam lithography and sputtering technique. Magneto-optic Kerr effect and ferromagnetic resonance analysis have confirmed the presence of dominant cubic and moderate uniaxial magnetic anisotropies in the thin films. Domain imaging via x-ray photoemission electron microscopy on the antidot arrays reveals chain like switching or correlated bigger domains for different shape of the antidots. Time-resolved MOKE microscopy has been performed to study the precessional dynamics and magnonic modes of the antidots with different shapes. We show that the optically induced spin-wave spectra in such antidot arrays can be tuned by changing the shape of the holes. The variation in internal field profiles, pinning energy barrier, and anisotropy modifies the spin-wave spectra dramatically within the antidot arrays with different shapes. We further show that by combining the magnetocrystalline anisotropy with the shape anisotropy, an extra degree of freedom can be achieved to control the magnonic modes in such antidot lattices.

cond-mat.mtrl-sci↗

arXiv2018

Hybrid magneto-dynamical modes in a single magnetostrictive nanomagnet on a piezoelectric substrate arising from magneto-elastic modulation of precessional dynamics

Magneto-elastic (or "straintronic") switching has emerged as an extremely energy-efficient mechanism for switching the magnetization of magnetostrictive nanomagnets in magnetic memory, logic and non-Boolean circuits. Here, we investigate the ultrafast magneto-dynamics associated with straintronic switching in a single quasi-elliptical magnetostrictive Co nanomagnet deposited on a piezoelectric PMN-PT substrate using time-resolved magneto-optical Kerr effect (TR-MOKE) measurements. The pulsed laser pump beam in the TR-MOKE plays a dual role: it causes precession of the nanomagnet's magnetization about an applied bias magnetic field and it also generates surface acoustic waves (SAWs) in the piezoelectric substrate that produce periodic strains in the magnetostrictive nanomagnet and modulate the precessional dynamics. This modulation gives rise to intriguing hybrid magneto-dynamical modes in the nanomagnet, with rich spin wave texture. The characteristic frequencies of these modes are 5-15 GHz, indicating that strain can affect magnetization in a magnetostrictive nanomagnet in time scales much smaller than 1 ns (~100 ps). This can enable ~10 GHz-range magneto-elastic nano-oscillators that are actuated by strain instead of a spin-polarized current, as well as ultrafast magneto-electric generation of spin waves for magnonic logic circuits, holograms, etc.

cond-mat.mes-hall↗