My Blog

Dimensions that matter

Resume

Technical Horoscope

GitHub

Codes


I am a Machine Learning Engineer at KLA.
I graduated with a Master's degree in Electrcical and Computer Engineering from the University of Michigan, Ann Arbor. My techincal intersts can be broadly classified as Machine Learning applications in Signal Processing and Data Analysis.

I have previously worked with Robert Laganiere in University of Ottawa on vehicle detection systems used for driver assistance and smart video surveillance applications. I worked towards my bachelor thesis under Dr. Prasanta Kumar Ghosh in Indian Institute of Science on analysis of the Indian Spoken English Pronunciation using rhythmic and prosodic cues.

Apart from this technical experience, I bring in with me a complete package of a 'Semi-classical singer', 'Convincing actor', 'Good orator' and above all, an 'Engaging Story-teller'. I like my audience big and friends circle small.

Publications

WACV 2020

CompressNet: Generative Compression at Extremely Low Bitrates

Vijayakrishna Naganoor, Shubham Dash, Giridharan Kumaravelu, Suraj Kiran Raman, Aditya Ramesh (code) (Paper)

IEEE-SITIS 2016

Selfie Detection by Synergy-Constriant Based Convolutional Neural Network

Yashas Annadani, Vijayakrishna Naganoor, Akshay Kumar Jagadish and Dr.Krishnan Chemmangat (Presentation) (code) (Paper)

TENCON 2016

Word Boundary Estimation for Continuous Speech Using Higher Order Statistical Features

Vijayakrishna Naganoor, Akshay Kumar Jagadish, and Dr.Krishnan Chemmangat (Presentation) (Paper)

Projects

Deep Learning Approach to Visual Question Answering

Vijayakrishna Naganoor, Suraj Kiran Raman, Shantakumar Venkataraman, Dr. Mert Pilanci (Paper)
Extracted Visual representation of images by fine-tuning ResNet and processed textual features using LSTM.
Developed D-VQA and DS-VQA models by integrating depth spatial features extracted using kinect depth map and achieved significant improvement in the WUPS score on the reduced DQUAR Dataset.

Second order optimizers for Deep Learning

Vijayakrishna Naganoor, Suraj Kiran Raman, Dr. Mert Pilanci
Worked on development of better second order optimisation methods in the place of the existing first order optimizers for faster convergence and better scaling for larger batch-size while training Deep Neural Networks.
Implemented Hessian free optimizers leveraging the idea of Gauss-Newton matrix and Conjugate Gradient methods

Contact Me

Email : vijaykn@umich.edu
Facebook