• CV
  
• Work Experience
  
• Publications
  
• Contact
  
• Email: {first_name}@cs.toronto.edu

Gagandeep Singh

/gəgəndip sɪŋgʰ/

Senior Research Scientist
Microsoft

Hello! I am a research scientist at Microsoft. I work on natural language processing for clinical documentation. This mostly involves the use of deep-learning for building sequence-to-sequence models. Before this I was a machine learning engineer at Yaar Inc., a Toronto based startup, where I worked on using machine learning for various applications in the domain of speech processing, which include speaker-verification, keyword-spotting and text-to-speech synthesis. During my M.Sc., I was a member of the Computational Linguistics group in the Department of Computer Science at the University of Toronto, my advisor being Prof. Gerald Penn. Before moving to Toronto, I obtained my Bachelor of Technology in Electronics and Communication Engineering at the Indian Institute of Technology Guwahati. Please refer my (not recently updated) CV for more information.

Research Interests

• Speech processing
• Machine learning
• Statistical modeling

Work Experience

Professional Experience

• Senior Research Scientist
I currently work on natural language processing for clinical documentation.
• Machine Learning Engineer
I worked in the speech team of Yaar Inc. Some of the applications that I worked on are speaker verification, keyword spotting and text-to-speech synthesis.
• Part-time Consultant
Worked with a Toronto based start-up to enhance the speech recognition component for an iOS application. The application used CMU sphinx library for implementing the ASR.

Research

• Research Assistant, Computational Linguistics group @ University of Toronto, Toronto, Canada.
My research area is statistical parametric speech synthesis with a focus on linear dynamical and autoregressive models, which are a class of continuous state space models. The project involves developing these statistical models and comparing their performance with deep neural network based models. research paper
• Research Assistant, Network Research Laboratory @ University of Toronto, Toronto, Canada.
Added components to Lotos, a simulation software tool, developed in Java for building overlay networks. The work included implementing CSMA-CA protocol for wireless nodes and adding option for generating multiple tracks for simulating wireless nodes installed on trains.
• Bachelor’s Thesis Project, EEE department @ IIT Guwahati, Guwahati, India.
Using a simple rateless encoding scheme, several opportunistic multicast algorithms were proposed and their performance studied in test networks using simulations. [thesis pdf] [publication]
• Summer Research Internship, Vodafone Chair Mobile Comm. Systems @ Technische Universität Dresden, Dresden, Germany.
The project involved development of methods for generation and detection of finite length reconstructible ASK sequences to be used in a communication system employing single bit quantization at the receiver in order to enhance performance of the ADC in multi-gigabit-per-second communications. [publication]
• Summer Research Internship, VNext Generation Wireless Systems Lab @ IISc Bangalore, Bangalore, India.
The project aimed to develop novel relay selection rules in a cognitive radio environment with the secondary system using amplify-forward relays. [publication]

Teaching Assistant

• Teaching Assistant, Advanced Engineering Mathematics
• Teaching Assistant, Electrical Fundamentals
• Teaching Assistant, Introduction to Computing

Publications

• G. Singh, Y. Pan, J. Andres-Ferrer, M. Del-Agua, F. Diehl, J. Pinto, P. Vozila, "Large Scale Sequence-to-Sequence Models for Clinical Note Generation from Patient-Doctor Conversations", Proceedings of the 5th Clinical Natural Language Processing Workshop, pages 138–143, Toronto, Canada. Association for Computational Linguistics. [pdf]
• G. Michalopoulos, K. Williams, G. Singh, T. Lin, "MedicalSum: A Guided Clinical Abstractive Summarization Model for Generating Medical Reports from Patient-Doctor Conversations", Findings of the Association for Computational Linguistics: EMNLP 2022 [pdf]
• A. Frydenlund, G. Singh, F. Rudzicz, "Language Modelling via Learning to Rank", Proceedings of the AAAI 2022 Conference on Artificial Intelligence [pdf]
• S. Enarvi, M. Amoia, M. Del-Agua Teba, B. Delaney, F. Diehl, S. Hahn, K. Harris, L. McGrath, Y. Pan, J. Pinto, L. Rubini, M. Ruiz, G. Singh, F. Stemmer, W. Sun, P. Vozila, T. Lin and R. Ramamurthy, "Generating Medical Reports from Patient-Doctor Conversations Using Sequence-to-Sequence Models", First Workshop on Natural Language Processing for Medical Conversations, July 2020, [pdf]
• G. Singh, "Speech Synthesis Using Linear Dynamical Models" (Master's thesis), University of Toronto, December 2017 [pdf]
• G. Singh, L. Landau and G. Fettweis, "Finite Length Reconstructible ASK-Sequences Received with 1-bit Quantization and Oversampling," SCC 2015; 10th International ITG Conference on Systems, Communications and Coding, Hamburg, Germany, 2015, pp. 1-6. [pdf]
• P. Das, N. B. Mehta and G. Singh, "Novel Relay Selection Rules for Average Interference-Constrained Cognitive AF Relay Networks," in IEEE Transactions on Wireless Communications, vol. 14, no. 8, pp. 4304-4315, Aug. 2015. [pdf]
• G. B. Shankar, G. Singh, S. Bose and W. D. Zhong, "Multicasting in wireless networks using rateless codes and opportunistic routing," 2015 10th International Conference on Information, Communications and Signal Processing (ICICS), Singapore, 2015, pp. 1-5.
[pdf]

Presentations

Statistical Parametric Speech Synthesis

Multicasting in Wireless Networks using Rateless Codes and Opportunistic Routing

Finite Length Reconstructible ASK-Sequences Received with 1-bit Quantization and Oversampling

Contact

[first_name] @ cs.toronto.edu
Find me on LinkedIn · GitHub

Template design by Andreas Viklund