IDDD 2023-2025 Batch

Sheth Janil Samir
B.Tech: Bioengineering
Masters: Complex Systems and Dynamics

BE20B031

Integrated model for sleep EEG reconstruction
Video

Advisor: Prof Srinivas Chakravarthy

In this work, we propose a generative oscillatory neural network for reconstruction of EEG signals across different sleep stages wake, nrem and rem. The proposed architecture consists of 3 layers : first being the oscillatory layer, 2nd a tanh neurons hidden layer and 3rd the output layer. Network is trained in 2 stages, firstly training the oscillators of the oscillatory layer by reconstructing the EEG signals and then the generative network trained with external activation to reconstruct the EEG. Our results indicate quite accurate reconstruction of the signals for all sleep stages.
Thesis

Placement FinMechanics

Aarohee Anand Vora
B.Tech: Bioscience
Masters: Complex Systems and Dynamics

BS20B002

Optimising Itaconic Acid Production on Acetate
Video

Advisor: Professor Himanshu Sinha and Professor Lars Blank

This work establishes a framework that integrates in silico metabolic predictions with targeted genetic engineering to improve itaconic acid production in U. maydis. The findings support future evaluation of predicted gene targets for strain improvement and demonstrate the viability of acetate as a sole substrate for efficient itaconate biosynthesis.
Thesis

Placement AstraZeneca

Kamesh P
B.Tech: Engineering Design
Masters: Complex Systems and Dynamics

ED20B027

Unified Memory Retention and Processing in Neural Networks
Video

Advisor: Srinivasa Chakravarthy

My DDP project explores the design and implementation of a memory-based neural network architecture that can simultaneously process it. By integrating flip-flop based neural architecture and the idea of list indexing based memory, the proposed model aims to achieve efficient, context(index)-dependent memory storage, retrieval and other processes on it.
Thesis

Placement PhD, Engineering Design, IIT Madras

Ujjwal Vishwanth KV
B.Tech: Mechanical Engineering
Masters: Complex Systems and Dynamics

ME20B194

ME20B194

Computational Frameworks for Modelling Large Ordered Complex Networked Dynamics
Video

Advisor: Sayan Gupta

I address solving large-scale coupled nonlinear equations in complex systems (e.g., epidemic dynamics, coupled oscillators, etc.). Due to lack of closed-form solutions, numerical methods have to be used. I propose hardware-dependent (CPU/GPU parallelism) and agnostic (analytical treatment: LRA, adj comp, precomputation, memory reduction, etc.) techniques. Applied to epidemic models on synthetic and real networks (C-elegans, Yeast, Charlottesville), methods are benchmarked across topologies and parameters. Massive speedups are observed, with methods catering to varied network topologies.
Thesis

Placement HSBC

Ishwarya Ganesh
B.Tech: Material Science
Masters: Complex Systems and Dynamics

MM20B025

Reservoir Computing and Swarm Dynamics
Video

Advisor: Sayan Gupta

This project explores predicting dynamics in systems inspired by the Boids model using a reservoir computing machine. Two cases are studied: (1) a swarm with a predator following a chaotic path, and (2) a swarm pursued by realistic predators. In both, the predator’s path is predicted from the swarm's collective motion using machine learning, enabling insight into predator dynamics without modeling their behavior directly. The goal is to advance data-driven prediction in complex, emergent systems.
Thesis

Placement FinMechanics

Aniket Malviya
B.Tech: Naval Architecture
Masters: Complex Systems and Dynamics

NA20B006

Desynchronisation of synchronised globally coupled Rossler Network using Deep Brain Stimulus(DBS).
Video

Advisor: Sayan Gupta

This thesis explores adaptive control strategies to suppress undesired synchrony in globally coupled oscillators, relevant to neurological disorders like Parkinson’s disease. Focusing on pulsatile stimulation at sensitive phases, a closed-loop scheme is developed for real-time phase estimation and adaptive tuning with minimal intervention. The method is tested on spiking and bursting neuron models. An open-loop framework is also examined for comparison, highlighting the superior efficiency and robustness of feedback-based control for neuromodulation.
Thesis

Placement  Project associate

Complex Systems & Dynamics     Indian Institute of Technology Madras     Chennai 600036     India