Applied Scientist II @ Amazon | Speech & NLP Expert | PhD @ IIT Madras

Posts by Collection

portfolio

publications

Spike Estimation from Fluorescence Signals Using High-Resolution Property of Group Delay

Published in IEEE Transactions on Signal Processing, 2019

While existing methods rely on data-driven methods and the physiology of neurons for modelling the spiking process, this work exploits the nature of the fluorescence responses to spikes using signal processing. We first motivate the problem by a novel analysis of the high-resolution property of minimum-phase group delay (GD) functions for multi-pole resonators. The resonators Read more

Subspace techniques for task independent EEG person identification

Published in In the proceedings of IEEE International Conference of Engineering in Medicine and Biology Society (EMBS), 2019

To effectively extract person-specific signatures present in EEG, it is necessary to define a subspace that enhances the biometric information and suppresses other nuisance factors. i-vector and x-vector are state-of-art subspace techniques used in speaker recognition. In this paper, novel modifications are proposed for both frameworks to project person-specific signatures from multi-channel EEG into a Read more

Time Warping Solutions for Classifying Artifacts in EEG

Published in In the proceedings of IEEE International Conference of Engineering in Medicine and Biology Society (EMBS), 2019

In this paper, we devise algorithms for detection and classification of artifacts. Classification of artifacts into head nod, jaw movement and eye-blink is performed using two different varieties of time warping; namely, linear time warping, and dynamic time warping. The average accuracy of 85% and 90% is obtained using the former, and the later, respectively. Read more

Spoof Detection Using Time-Delay Shallow Neural Network and Feature Switching

Published in In the proceedings of 2019 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), 2019

Inspired by the state-of-the-art x-vector based speaker verification approach, this paper proposes a time-delay shallow neural network (TD-SNN) for spoof detection for both logical and physical access. The novelty of the proposed TD-SNN system vis-a-vis conventional DNN systems is that it can handle variable length utterances during testing. Performance of the proposed TD-SNN systems and Read more

Seizure Detection Using Time Delay Neural Networks and LSTMs

Published in In the proceedings of 2020 IEEE Signal Processing in Medicine and Biology Symposium (SPMB), 2020

In this paper, we propose a neural network system using the time-delay neural network to model temporal information (TDNN) and long short term memory (LSTM) layer to model spatial information. On the development subset of Temple University seizure dataset, the proposed system achieved a sensitivity of 23.32 % with 11.13 false alarms in 24 hours. Read more

Evidence of Task-Independent Person-Specific Signatures in EEG using Subspace Techniques

Published in IEEE Transactions on Information Forensics and Security, 2021

The paper extends ideas from subspace-based text-independent speaker recognition and proposes novel modifications for modeling multi-channel EEG data. The proposed techniques assume that biometric information is present in the entire EEG signal and accumulate statistics across time in a high dimensional space. These high dimensional statistics are then projected to a lower dimensional space where Read more

Functional parcellation of mouse visual cortex using statistical techniques reveals response-dependent clustering of cortical processing areas

Published in PLOS Computational Biology, 2021

In contrast to the conventional input-output characterizations of neuronal responses to standard visual stimuli, here we asked whether six of the core visual areas have responses that are functionally distinct from each other for a given visual stimulus set, by applying machine learning techniques to distinguish the areas based on their activity patterns. Visual areas Read more

Dual Script E2E Framework for Multilingual and Code-Switching ASR

Published in In the proceedings of Proc. Interspeech, 2021

In this paper, we use an in-house rule-based phoneme-level common label set (CLS) representation to train multilingual and code-switching ASR for Indian languages. We propose a modification to the E2E model, wherein the CLS representation and the native language characters are used simultaneously for training. We show our results on the multilingual and code-switching (MUCS) Read more

talks

Task-Independent EEG based Subject Identification using Auditory Stimulus

Published:

Recent studies have shown that task-specific electroencephalography (EEG) can be used as a reliable biometric. This paper extends this study to task-independent EEG with auditory stimuli. Data collected from 40 subjects in response to various types of audio stimuli, using a 128 channel EEG system is presented to different classifiers, namely, k-nearest neighbour (k-NN), artificial neural network (ANN) and universal background model - Gaussian mixture model (UBM-GMM). It is observed that k-NN and ANN perform well when testing is performed intra-session, while UBM-GMM framework is more robust when testing is performed intersession. This can be attributed to the fact that the correspondence of the sensor locations across sessions is only approximate. It is also observed that EEG from parietal and temporal regions contain more subject information although the performance using all the 128 channel data is marginally better. Read more

Subspace techniques for task independent EEG person identification

Published:

There has been a growing interest in studying electroencephalography signals (EEG) as a possible biometric. The brain signals captured by EEG are rich and carry information related to the individual, tasks being performed, mental state, and other channel/measurement noise due to session variability and artifacts. To effectively extract person-specific signatures present in EEG, it is necessary to define a subspace that enhances the biometric information and suppresses other nuisance factors. i-vector and x-vector are state-of-art subspace techniques used in speaker recognition. In this paper, novel modifications are proposed for both frameworks to project person-specific signatures from multi-channel EEG into a subspace. The modified i-vector and x-vector systems outperform baseline i-vector and x-vector systems with an absolute improvement of 10.5% and 15.9%, respectively Read more

EEG signals and task-independent person-specific signatures

Published:

Recent studies have shown interest in using brain signals collected through electroencephalography (EEG) as a reliable biometric. However, owing to limited data, the focus of the field has been limited to classical techniques using different elicitation protocols. The primary interest for studying different elicitation protocols is that individuals will have a distinctive signature for a given task which can be leveraged to identify them. We conjecture that the biometric signatures should be present in the EEG signal irrespective of tasks or state of the brain. Both speech and brain signals are temporal data which can be described as a sequence of observations; the analysis of these signals to get the required information has always been a challenging task. In this talk, we draw parallels between the problem of biometric recognition using speech and brain signals. For decades, owing to the enormous amount of data available, a lot of machine learning and statistical techniques have been proposed to extract biometric information from a speech signal irrespective of the spoken content. Using the universal background model - Gaussian mixture model (UBM-GMM), a text-independent speaker verification technique, we first verify the conjecture that biometric signatures should be present in the EEG signal independent of task. Later, we extend and modify the current state-of-art subspace-based speaker verification techniques to identify individuals from EEG signals irrespective of the task reliably. Read more

teaching

Teaching experience 1

Undergraduate course, University 1, Department, 2014

This is a description of a teaching experience. You can use markdown like any other post. Read more

Teaching experience 2

Workshop, University 1, Department, 2015

This is a description of a teaching experience. You can use markdown like any other post. Read more