collaborators

7 papers

cs.LG2026

Agile Reinforcement Learning through Separable Neural Architecture and Applications

Rajib Mostakim, Reza T. Batley, Sourav Saha

Deep reinforcement learning (RL) is increasingly deployed in resource-constrained environments, yet go-to function approximators - multilayer perceptrons (MLPs) - are often paramet…

cs.LG2026

Separable Neural Architectures as Physical World Models: from Mathematical Theory to Applications

Reza T Batley, Andrew Kichline, Sourav Saha

This work introduces the Separable Neural Architecture (SNA), a function representational class combining neural approximation with tensor decomposition. The SNA decouples localize…

cs.CL2026

Leviathan: Decoupling Input and Output Representations in Language Models

Reza T. Batley, Sourav Saha

Modern language models use a single matrix for input embedding and output projection. This couples two distinct objectives: token representation and discrimination over a vocabular…

cs.LG2026

Separable neural architectures as a primitive for unified predictive and generative intelligence

Reza T. Batley, Apurba Sarker, Rajib Mostakim +2

Intelligent systems across physics, language and perception often exhibit factorisable structure, yet are typically modelled by monolithic neural architectures that do not explicit…

cs.LG2025

A Unified Generative-Predictive Framework for Deterministic Inverse Design

Reza T. Batley, Sourav Saha

Inverse design of heterogeneous material microstructures is a fundamentally ill-posed and famously computationally expensive problem. This is exacerbated by the high-dimensional de…

cs.LG2025

The Method of Infinite Descent

Reza T. Batley, Sourav Saha

Training - the optimisation of complex models - is traditionally performed through small, local, iterative updates [D. E. Rumelhart, G. E. Hinton, R. J. Williams, Nature 323, 533-5…