Scholar
Rupesh Kumar Srivastava
Google Scholar ID: vTWuk1gAAAAJ
NNAISENSE
Citations & Impact
Citations
11,739
H-index
15
i10-index
22
Publications
20
Co-authors
36
list available
Publications
Resume (English only)
Academic Achievements
- Highlights of his past research include Bayesian Flow Networks: a new generative modeling framework that combines Bayesian inference with Deep Learning, and gracefully extends the ideas behind diffusion models to discrete data; Upside-Down Reinforcement Learning: a new way of learning to act from reinforcement that avoids reward prediction in favor of relying purely on supervised learning; Highway Networks/Recurrent Highway Networks: the first neural network architecture that enabled training networks with very large depths of tens to hundreds of layers, this was a predecessor and general version of now-common Residual Networks.
Research Experience
- Currently at the Institute of Foundation Models in Silicon Valley, leading the new Agents team and working on Reinforcement Learning and Reasoning research. Previously, he was a Senior Research Scientist at NNAISENSE, one of the first employees at the company, and also managed the software infrastructure team. During an internship at Microsoft Research in 2015, he was part of the team that developed and published one of the first neural networks based image captioning systems, tying (with Google) for the First place at the COCO Image Captioning Challenge in 2015.
Education
- Completed his PhD in 2018 at the Swiss AI lab IDSIA/USI in Lugano, Switzerland, supervised by Jürgen Schmidhuber. His dissertation was on New Architectures for Very Deep Learning, focused on the training of very deep networks. Obtained a Bachelors and Masters in Mechanical Engineering from IIT Kanpur, with a Masters thesis on using evolutionary algorithms for reliable design optimization, supervised by Kalyanmoy Deb.
Background
- An Artificial Intelligence researcher, interested in learning algorithms (algorithms for learning from data) and the idea that instead of designing algorithms, we should be learning them. Believes that learning is fundamentally about compression, and thinking about compression is likely the most promising way of making progress towards better and more general learning algorithms.
Miscellany
- Personal interests not mentioned.
