Sotabase

Career

· Researcher, UC Berkeley Berkeley Artificial Intelligence Research Lab (BAIR)2024–
· Researcher (Thesis/dissertation), UC Berkeley2019–
· Graduate Student in Electrical Engineering and Computer Sciences, UC Berkeley2006–
· Teaching Assistant/Instructor, UC Berkeley2002–

Publications (13)

Cross-Entropy Loss and Low-Rank Features Have Responsibility for Adversarial Examples
arXiv.org · 2019
59
cited
Step Size Matters in Deep Learning
Neural Information Processing Systems · 2018
47
cited
Sampling multidimensional Wiener processes
IEEE Conference on Decision and Control · 2014
42
cited
Persistency of Excitation for Robustness of Neural Networks
arXiv.org · 2019
13
cited
Cross-Entropy Loss Leads To Poor Margins
2018
12
cited
Learning prospect theory value function and reference point of a sequential decision maker
IEEE Conference on Decision and Control · 2017
9
cited
An Analytical Framework to Address the Data Exfiltration of Advanced Persistent Threats
IEEE Conference on Decision and Control · 2018
3
cited
Residual Networks: Lyapunov Stability and Convex Decomposition
arXiv.org · 2018
3
cited
Learning Unstable Dynamical Systems with Time-Weighted Logarithmic Loss
arXiv.org · 2020
2
cited
Opportunistic sensor activation in the face of data deluge
IEEE Conference on Decision and Control · 2013
Optimal sampling and switching policies for linear stochastic systems
2014
Optimization, Robustness and Risk-Sensitivity in Machine Learning: A Dynamical Systems Perspective
2020
Richness of Training Data Does Not Suffice: Robustness of Neural Networks Requires Richness of Hidden-Layer Activations
2020
Sotabase