This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
sdmia_invited_speakers [2015/09/25 18:53]
sdmia_invited_speakers [2015/11/13 12:07] (current)
Line 24: Line 24:
 [[http://​www.cs.cmu.edu/​~ebrun/​|CMU]] [[http://​www.cs.cmu.edu/​~ebrun/​|CMU]]
-Title: ​TBD\\+Title: ​**Quickly Learning to Make Good Decisions**\\
 Abstract:\\ Abstract:\\
-TBD+A fundamental goal of artificial intelligence is to create agents that 
 +learn to make good decisions as they interact with a stochastic 
 +environment. Some of the most exciting and valuable potential 
 +applications involve systems that interact directly with humans, such as 
 +intelligent tutoring systems or medical interfaces. In these cases, 
 +sample efficiency is highly important, as each decision, good or bad, 
 +is impacting a real person. I will describe our research on tackling 
 +this challenge, as well as its relevance to improving educational tools. 
 === Alan Fern === === Alan Fern ===
 [[http://​web.engr.oregonstate.edu/​~afern/​|Oregon State]] [[http://​web.engr.oregonstate.edu/​~afern/​|Oregon State]]
-Title: ​TBD\\+Title: ​**Kinder and Gentler Teaching Modes for Human-Assisted Policy Learning**\\
 Abstract:\\ Abstract:\\
-TBD+This talk considers the problem of teaching action policies to computers for sequential decision making. The vast majority of policy learning algorithms offer human teachers little flexibility in how policies are taught. In particular, one of two learning modes is typically considered: 1) Imitation learning, where the teacher demonstrates explicit action sequences to the learner, and 2) Reinforcement learning, where the teacher designs a reward function for the learner to autonomously optimize via practice. This is in sharp contrast to how humans teach other humans, where many other learning modes are commonly used besides imitation and practice. The talk will highlight some of our recent work on broadening the available learning modes for computer policy learners, with the eventual aim of allowing humans to teach computers more naturally and efficiently. In addition, we will sketch some of the challenges in this research direction for both policy learning and more general planning systems. 
 === Mykel Kochenderfer === === Mykel Kochenderfer ===
Line 45: Line 55:
 [[http://​teamcore.usc.edu/​tambe/​|USC]] [[http://​teamcore.usc.edu/​tambe/​|USC]]
-Title: ​TBD\\+Joint work with Eric Rice, Amulya Yadav, and Robin Petering. 
 +Title: ​**PSINET: Assisting HIV Prevention Amongst Homeless Youth using POMDPs**\\
 Abstract:\\ Abstract:\\
-TBD+Homeless youth are prone to Human Immunodeficiency 
 +Virus (HIV) due to their engagement in high risk behavior 
 +such as unprotected sex, sex under influence of 
 +drugs, etc. Many non-profit agencies conduct interventions 
 +to educate and train a select group of homeless 
 +youth about HIV prevention and treatment practices and 
 +rely on word-of-mouth spread of information through 
 +their social network. Previous work in strategic selection 
 +of intervention participants does not handle uncertainties 
 +in the social network’s structure and evolving 
 +network state, potentially causing significant shortcomings 
 +in spread of information. Thus, we developed 
 +PSINET, a decision support system to aid the agencies 
 +in this task. PSINET includes the following key novelties:​ 
 +(i) it handles uncertainties in network structure 
 +and evolving network state; (ii) it addresses these uncertainties 
 +by using POMDPs in influence maximization;​ 
 +and (iii) it provides algorithmic advances to allow high 
 +quality approximate solutions for such POMDPs. We are about 
 +to conduct a pilot test study with homeless youth in Los Angeles; 
 +we will present a progress report. ​
 === Jason Williams === === Jason Williams ===
Line 59: Line 91:
 [[http://​rbr.cs.umass.edu/​shlomo/​|UMass Amherst]] [[http://​rbr.cs.umass.edu/​shlomo/​|UMass Amherst]]
-Title: ​TBD\\+Title: ​**Do We Expect Too Much from DEC-POMDP Algorithms?​**\\
 Abstract:\\ Abstract:\\
-TBD+Sequential decision models such as DEC-POMDPs are powerful and elegant approaches for planning in situations that involve multiple cooperating decision makers. They are powerful in the sense that we can, in principle, capture a rich class of problems. ​ They are elegant in the sense that they include the minimal set of ingredients needed to analyze these problems and facilitate rigorous mathematical examination of their fundamental properties. ​ An optimal solution of a DEC-POMDP explicitly answers the question of what should an agent do to maximize value. ​ Implicitly, an optimal solution answers many other questions including the appropriate assignment of meaning to internal memory states, appropriate adoption of goals and subgoals, appropriate assignment of roles to agents, and appropriate assignment of meaning to messages that agents exchange. In fact, an optimal policy optimizes all these choices implicitly. In this talk, I argue that this is just too much to expect from a computational point of view.  There is much to be gained by decomposing the planning problem in a way that some of these questions are answered first and a simplified planning problem is then solved. ​ I discuss a few examples of such decompositions and examine their contribution to the scalability of planning algorithms
Recent changes RSS feed Creative Commons License Donate Minima Template by Wikidesign Driven by DokuWiki