Zobrazeno 1 - 10
of 4 927
pro vyhledávání: '"Moitra A."'
Motivated by connections between algebraic complexity lower bounds and tensor decompositions, we investigate Koszul-Young flattenings, which are the main ingredient in recent lower bounds for matrix multiplication. Based on this tool we give a new al
Externí odkaz:
http://arxiv.org/abs/2411.14344
Autor:
Liu, Allen, Moitra, Ankur
Model stealing, where a learner tries to recover an unknown model via carefully chosen queries, is a critical problem in machine learning, as it threatens the security of proprietary models and the privacy of data they are trained on. In recent years
Externí odkaz:
http://arxiv.org/abs/2411.07536
Autor:
Rohatgi, Dhruv, Marwah, Tanya, Lipton, Zachary Chase, Lu, Jianfeng, Moitra, Ankur, Risteski, Andrej
Graph neural networks (GNNs) are the dominant approach to solving machine learning problems defined over graphs. Despite much theoretical and empirical work in recent years, our understanding of finer-grained aspects of architectural design for GNNs
Externí odkaz:
http://arxiv.org/abs/2410.09867
We consider the problem of learning graphical models, also known as Markov random fields (MRFs) from temporally correlated samples. As in many traditional statistical settings, fundamental results in the area all assume independent samples from the d
Externí odkaz:
http://arxiv.org/abs/2409.05284
Publikováno v:
Applied Physics Reviews, 2024
This review explores the intersection of bio-plausible artificial intelligence in the form of Spiking Neural Networks (SNNs) with the analog In-Memory Computing (IMC) domain, highlighting their collective potential for low-power edge computing enviro
Externí odkaz:
http://arxiv.org/abs/2408.12767
Due to the high computation overhead of Vision Transformers (ViTs), In-memory Computing architectures are being researched towards energy-efficient deployment in edge-computing scenarios. Prior works have proposed efficient algorithm-hardware co-desi
Externí odkaz:
http://arxiv.org/abs/2408.12742
Autor:
Moitra, Upamanyu
We consider trajectories of massless particles in the presence of charged black holes in asymptotically AdS spacetimes in arbitrary dimensions. We study the properties of the photon ring in the (near-)extremal limit and show that the photon ring can
Externí odkaz:
http://arxiv.org/abs/2408.08308
Autor:
Dabholkar, Atish, Moitra, Upamanyu
Publikováno v:
JHEP 12 (2024) 012
We construct $\mathbb{Z}_N$ orbifolds of the ten-dimensional heterotic string theories appropriate for implementing the stringy replica method for the calculation of quantum entanglement entropy. A novel feature for the heterotic string is that the g
Externí odkaz:
http://arxiv.org/abs/2407.17553
Autor:
Golowich, Noah, Moitra, Ankur
In this paper, we study the offline RL problem with linear function approximation. Our main structural assumption is that the MDP has low inherent Bellman error, which stipulates that linear value functions have linear Bellman backups with respect to
Externí odkaz:
http://arxiv.org/abs/2406.11686
Autor:
Golowich, Noah, Moitra, Ankur
One of the most natural approaches to reinforcement learning (RL) with function approximation is value iteration, which inductively generates approximations to the optimal value function by solving a sequence of regression problems. To ensure the suc
Externí odkaz:
http://arxiv.org/abs/2406.11640