Picture for Finale Doshi-Velez

Finale Doshi-Velez

Model-based Reinforcement Learning for Semi-Markov Decision Processes with Neural ODEs

Jun 29, 2020
Jianzhun Du, Joseph Futoma, Finale Doshi-Velez

Figure 1 for Model-based Reinforcement Learning for Semi-Markov Decision Processes with Neural ODEs

Figure 2 for Model-based Reinforcement Learning for Semi-Markov Decision Processes with Neural ODEs

Figure 3 for Model-based Reinforcement Learning for Semi-Markov Decision Processes with Neural ODEs

Figure 4 for Model-based Reinforcement Learning for Semi-Markov Decision Processes with Neural ODEs

* 20 pages, 7 figures

Access Paper or Ask Questions

Learned Uncertainty-Aware (LUNA) Bases for Bayesian Regression using Multi-Headed Auxiliary Networks

Jun 21, 2020
Sujay Thakur, Cooper Lorsung, Yaniv Yacoby, Finale Doshi-Velez, Weiwei Pan

Figure 1 for Learned Uncertainty-Aware (LUNA) Bases for Bayesian Regression using Multi-Headed Auxiliary Networks

Figure 2 for Learned Uncertainty-Aware (LUNA) Bases for Bayesian Regression using Multi-Headed Auxiliary Networks

Figure 3 for Learned Uncertainty-Aware (LUNA) Bases for Bayesian Regression using Multi-Headed Auxiliary Networks

Figure 4 for Learned Uncertainty-Aware (LUNA) Bases for Bayesian Regression using Multi-Headed Auxiliary Networks

* 17 pages, 15 figures, arXiv:1412.6980, arXiv:1904.00577

Access Paper or Ask Questions

PAC Bounds for Imitation and Model-based Batch Learning of Contextual Markov Decision Processes

Jun 11, 2020
Yash Nair, Finale Doshi-Velez

Access Paper or Ask Questions

Characterizing and Avoiding Problematic Global Optima of Variational Autoencoders

Mar 17, 2020
Yaniv Yacoby, Weiwei Pan, Finale Doshi-Velez

Figure 1 for Characterizing and Avoiding Problematic Global Optima of Variational Autoencoders

Figure 2 for Characterizing and Avoiding Problematic Global Optima of Variational Autoencoders

Figure 3 for Characterizing and Avoiding Problematic Global Optima of Variational Autoencoders

Figure 4 for Characterizing and Avoiding Problematic Global Optima of Variational Autoencoders

* PMLR 118:1-17, 2020
* Accepted at the Proceedings of The 2nd Symposium on Advances in Approximate Bayesian Inference 2019

Access Paper or Ask Questions

Interpretable Off-Policy Evaluation in Reinforcement Learning by Highlighting Influential Transitions

Feb 14, 2020
Omer Gottesman, Joseph Futoma, Yao Liu, Sonali Parbhoo, Leo Anthony Celi, Emma Brunskill, Finale Doshi-Velez

Figure 1 for Interpretable Off-Policy Evaluation in Reinforcement Learning by Highlighting Influential Transitions

Figure 2 for Interpretable Off-Policy Evaluation in Reinforcement Learning by Highlighting Influential Transitions

Figure 3 for Interpretable Off-Policy Evaluation in Reinforcement Learning by Highlighting Influential Transitions

Figure 4 for Interpretable Off-Policy Evaluation in Reinforcement Learning by Highlighting Influential Transitions

* Change: Correction of typo in meta-data author names

Access Paper or Ask Questions

POPCORN: Partially Observed Prediction COnstrained ReiNforcement Learning

Jan 13, 2020
Joseph Futoma, Michael C. Hughes, Finale Doshi-Velez

Figure 1 for POPCORN: Partially Observed Prediction COnstrained ReiNforcement Learning

Figure 2 for POPCORN: Partially Observed Prediction COnstrained ReiNforcement Learning

Figure 3 for POPCORN: Partially Observed Prediction COnstrained ReiNforcement Learning

Figure 4 for POPCORN: Partially Observed Prediction COnstrained ReiNforcement Learning

* Accepted, to appear at AISTATS 2020, Palermo. Note that this version is not the final camera-ready; that will appear in a few weeks

Access Paper or Ask Questions

Identifying Distinct, Effective Treatments for Acute Hypotension with SODA-RL: Safely Optimized Diverse Accurate Reinforcement Learning

Jan 09, 2020
Joseph Futoma, Muhammad A. Masood, Finale Doshi-Velez

Figure 1 for Identifying Distinct, Effective Treatments for Acute Hypotension with SODA-RL: Safely Optimized Diverse Accurate Reinforcement Learning

Figure 2 for Identifying Distinct, Effective Treatments for Acute Hypotension with SODA-RL: Safely Optimized Diverse Accurate Reinforcement Learning

Figure 3 for Identifying Distinct, Effective Treatments for Acute Hypotension with SODA-RL: Safely Optimized Diverse Accurate Reinforcement Learning

Figure 4 for Identifying Distinct, Effective Treatments for Acute Hypotension with SODA-RL: Safely Optimized Diverse Accurate Reinforcement Learning

* Accepted for publication at the AMIA 2020 Informatics Summit. This version contains an updated appendix with additional figures not found in the page-constrained AMIA version, so treat this version as the most up-to-date

Access Paper or Ask Questions