Daniel Filan
Scholar

Daniel Filan

Google Scholar ID: 9eoaiXMAAAAJ
PhD Student, UC Berkeley
Artificial Intelligence
Citations & Impact
All-time
Citations
168
 
H-index
7
 
i10-index
7
 
Publications
18
 
Co-authors
9
list available
Resume (English only)
Academic Achievements
  • "Clusterability in Neural Networks": Introduced graph clustering to measure neuron grouping structure; showed networks trained with pruning/dropout are more clusterable.
  • "Exploring Hierarchy-Aware Inverse Reinforcement Learning": Presented at GoalsRL 2018; advocated hierarchical human models in IRL.
  • Co-authored web book "Modeling Agents with Probabilistic Programs" on agent modeling using WebPPL.
  • "Self-Modification of Policy and Utility Function in Rational Agents": Won the Kurzweil Prize for best paper at AGI 2016.
  • "Loss Bounds and Time Complexity for Speed Priors": Presented at AISTATS 2016, published in JMLR; analyzed time-sensitive algorithmic priors.