DeepDynamics Lab
Machine Learning × Dynamical Systems

DeepDynamics Lab

Machine intelligence through dynamics, geometry, and memory.

We study principled ways to build adaptive intelligent systems by combining dynamical systems, geometric structure, recurrent computation, and structured memory.

Led by Dr. Pradeep Singh at the Indian Institute of Information Technology Surat.

Research Publications People PI homepage

Research

Our work asks how dynamics and structure can become first-class computational primitives rather than afterthoughts.

  • 01

    Dynamics of Intelligence

    Recurrent computation, reservoir systems, temporal dynamics, structured memory, and emergent representations.

  • 02

    Geometry & Structure in Learning

    Hyperbolic representations, topology, equivariance, symmetry, and geometric inductive biases for generalisation.

  • 03

    Agentic & Adaptive Systems

    Memory, interaction, reasoning, and dynamical mechanisms for autonomous and continually adapting systems.

Research directions

Selected work

  • 2026

    What Reservoirs Can Say? Sofic Languages and Finite-Type Views of Echo State Dynamics

    P. Singh and R. Balasubramanian · IEEE Transactions on Neural Networks and Learning Systems · In press
  • 2026

    Holonomy Grid Codes for Generalisation Under Directed Actions

    P. Singh and R. Balasubramanian · Forty-Third International Conference on Machine Learning (ICML), Seoul
  • 2026

    Thalamocortical Burst Gating Enables Event-Locked Memory

    P. Singh and R. Balasubramanian · Knowledge-Based Systems
Selected publications