Natural Option Critic
Saket Tiwari, & Philip Thomas. (2019). "Natural Option Critic." AAAI 2019
CV
Education
- B.Tech. in Computer Science, IIT Bombay, 2014
- M.S. in Computer Science, University of Massachusetts, Amherst, 2019
- Ph.D in Computer Science, Brown University, 2024 (expected)
Work experience
- Summer 2018: Data Science Intern
- Researched the causal effects of user connections in social networks using deep variational auto-encoders for a large data set with millions of connections
- Selected as top 5 master’s students at UMass Amherst, from all the applicants.
- August 2016 - July 2017: Senior Machine Learning Engineer at LinkedIn
- Led the video classification project. Trained Deep Learning models and deployed them to production to handle more than 1000 fps, using asynchronous processing on GPU clusters
- Researched the training of Hierarchical Neural Auto-Encoders by introducing Neural Machine Translation type architecture and Tensorflow. Achieved a speed up of 2x
- Experimented with multiple implementations of a popularity based model, to achieve significant improvements in user engagement. Applied a Thompson sampling based explore-exploit model to engage cold users with companies
- July 2014 - July 2016: Machine Learning Engineer at LinkedIn
- Developed and maintained a high QPS service, which classified ALL the content created on LinkedIn (millions of entities every day) to filter out spam. Designed various parts of the system: training pipelines for machine learning models, deployment, synchronous and asynchronous classification
- Worked on a Hackday project to auto-classify advertisements posted on LinkedIn using SVM. The work lead to reduced wait time for ads to go live on LinkedIn.
Publications
ON THE GEOMETRY OF REINFORCEMENT LEARNING IN CONTINUOUS STATE AND ACTION SPACES
Saket Tiwari, Omer Gottesman, & George Konidaris. (2022). "On the Geometry of Reinforcement Learning in continuous state and action sapces." RLDM 2022
Effects of Data Geometry in Early Deep Learning
Saket Tiwari, & George Konidaris. (2022). "Effects of Data Geometry in Early Deep Learning." NeurIPS 2022
Skills
- Skill 1
- Skill 2
- Sub-skill 2.1
- Sub-skill 2.2
- Sub-skill 2.3
- Skill 3