Projects

• In my free time, I discuss ML with my undergraduate alma matter students. They can be found here: Link



• My Graduate Research:

  A deep learning framework for increasing the correlation between natural audio stimuli and the corresponding EEG signals. Thesis : pdf

  1. We successfully implemented a Deep version of Linear CCA methods proposed by Alain de Cheviegné et. al. It has been recently published at EMBC 2020. pdf

  2. We successfully proposed a Deep Multiway CCA framework for audio-EEG datasets of single-trial analysis. It has been published at ICASSP 2021. link

  3. We studied the deep learning framework for inter- and intra-subject audio-EEG analysis for audio and speech stimuli. It has been published at IEE TNSRE 2021. link

  4. Finally, we tried to reconstruct the speech stimuli from the recorded EEG responses. We tried to utilize the temporal context using LSTMs and transformers with attention, model the noise using generative adversarial loss, but couldn't succeed. More details are available in the thesis attached above.



• Course Projects:

  1. Sparse Signal Processing :

     Change-Point Detection and Estimation of Peicewise Constant Parameters using Sparse Linear Regression. pdf

  2. Speech Information Processing :

     Tried Spectral Offline Clustering for Speaker Diarization with LSTM. Link

  3. Data Analysis and Visualization :

     What single statistic exactly correlates with crime rate in India? pdf

• B. Tech Project: Smart Automotives

  An IoT project based on MSP430 to continuously assess the automobiles. Uploads data to cloud. Process the data in the cloud. Sends distress message if something unusual is reported. Link

• Fun (Side) Projects:

  1. Smart Home Automation :

     MSP430 based Low power automatic fire and intruder alert system. Provides an interface to control lights and fans.

  2. Internship Projects :

     1. A telepresence robot that can be controlled by an arduino device or bluetooth.
     2. A coin based charging point similar to rupee coin telephone boxes. Charges phones for a 5 minutes for 5 rupees.

  3. Neuromatch Academy Project:

     We worked on "Biologically inspired learning and optimization" problem. It was a one-week project, and we deployed our experiences from multiple fields to understand the problem. pdf