Michael Zavlanos
@m_zavlanos
Followers
238
Following
45
Statuses
21
Yoh Family Professor at @DukeEngineering
Durham, NC
Joined June 2020
RT @Boyuan__Chen: We are looking for two robotics/autonomy faculties here at Duke University! Please help spread the word and apply! We al…
0
6
0
RT @Boyuan__Chen: Excited to share our latest work on "Policy Stitching” published on CoRL 2023 this week! We focus on how to enable cross-…
0
8
0
RT @DukeAIHealth: Duke AI Health Director @PencinaPhd and Director of Governance and Evaluation of Health AI Systems, Nicoleta Economou, Ph…
0
1
0
RT @hansmriess: I'm excited to share our new paper w/ @mungowitz and @m_zavlanos introducing a decentralized mechanism for pricing and exch…
0
5
0
RT @DukeAIHealth: ICYMI: @DukeAIHealth's Nicoleta J. Economou, PhD, the founding director of the Algorithm-Based Clinical Decision Support…
0
2
0
RT @Boyuan__Chen: Our new lab at Duke University is looking for PhD students! We conduct research in robotics (both mechanical and computat…
0
60
0
Exciting new postdoc opportunity @DukeU on ethical AI, at the intersection of engineering, law, and policy. For more information and to apply visit:
0
7
10
RT @arXiv_Daily: Risk-Averse Multi-Armed Bandits with Unobserved Confounders: A Case Study in Emotion Regulation in Mobile Health https://t…
0
2
0
Thrilled to share the news of my promotion to Full Professor @DukeU! I am grateful to all the amazing students (present and past), collaborators, colleagues, mentors, family and friends who have inspired me and supported me along the way. Thank you!
12
1
103
New preprint on risk-averse learning in online convex games: We address important challenges related to the estimation of the risk values and gradients using finite feedback.
1
0
7
Two postdoc positions @Duke on optimization and machine learning. Projects span robotics, cyber-physical systems, and healthcare/medicine (with Biostatistics & Bioinformatics). To learn about our research, visit To apply, please e-mail me with a CV.
0
3
4
Learning neural network controllers with closed-loop safety guarantees requires a lot of data. But what if these data are not available? Check out our new preprint on learning neural network controllers that can fail with grace:
1
0
4
How can one learn optimal policies in continuous space from demonstrator data with hidden contextual information? Check out our recent work on continuous transfer RL with unobserved context presented at #L4DC. Paper: Talk (min 53):
0
0
3
Duke MEMS is hiring! If you are looking for a faculty position, please apply!
0
0
0
New postdoctoral position on machine learning and AI at Duke University, with focus on algorithms as well as applications in robotics, cyber-physical systems, or healthcare/medicine. For more information and to apply, please visit the link:
0
3
6
Residual feedback can also be used to efficiently estimate local policy gradients in distributed reinforcement learning, using local estimates of the global accumulated rewards that depend on partial state and action information only:
0
1
0