Portrait of Ece Kamar

Ece Kamar

Researcher

About

I am a researcher at the Adaptive Systems and Interaction group at Microsoft Research Redmond. I received a Ph.D. in computer science from Harvard University in 2010. My thesis focused on reasoning under uncertainty for successful human-computer teamwork. I received a M.S. from Harvard University in 2007 and a B.S. from Sabanci University in Turkey in 2005.

I work on several subfields of AI; including planning, machine learning, multi-agent systems and human-computer teamwork. I particularly focus on real-world applications that can benefit from the complementary abilities of humans and machines.

I served as a member of the first study panel of AI 100. Our report is available here.

To learn more, please visit my personal website here.

Announcements

  • I will be attending AAAI 2017. I will be giving a tutorial with Matthew Taylor and Brad Hayes on Interactive Machine Learning. I will be presenting at the Crowdsourcing, Deep Learning, and Artificial Intelligence Agents Workshop and at the Increasing Diversity in AI Workshop. We also have two technical papers to present.
  • I will be visiting University of Michigan on March 20th.

Projects

Microtasks and Microproductivity

Established: January 1, 2015

As people's time becomes increasingly fragmented, getting tasks accomplished becomes more and more challenging. In this project we are investigating ways of decomposing larger tasks into smaller microtasks that leverage the micromoments that exist sporadically during a person's day. Our…

IntelliDrive – Intelligent Management of Attention in Automotive Environments

Established: January 1, 2009

Driver attention is a valuable commodity in maintaining driving safety. However, with the proliferation of many interactive devices that place demands on the driver's attention while driving, effectively allocating attention with the primary goal of managing driving safety presents substantial…

Dialog and Conversational Systems Research

Established: March 14, 2014

Conversational systems interact with people through language to assist, enable, or entertain. Research at Microsoft spans dialogs that use language exclusively, or in conjunctions with additional modalities like gesture; where language is spoken or in text; and in a variety…

Publications

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

Projects

Other

Professional Activities

  • Participant of First Study Panel of One Hundred Year Study on Artificial Intelligence
  • Program Co-chair, WIP Track, HCOMP 2016, 2017
  • Program Committee Member, Collective Intelligence 2017
  • Program Committee Member, Workshop on Crowdsourcing, Deep Learning and Artificial Intelligence Agents 2017
  • Area Chair, IJCAI 2016
  • Senior Program Committee Member, AAMAS 2014, 2015, 2016
  • Organizer, Workshop on Human-Agent Interaction Design and Models at AAMAS, 2015; at IJCAI 2016
  • Program Committee Member, AAAI 2013, 2014, 2016, 2017
  • Program Committee Member, WWW, 2014
  • Program Committee Member, UAI 2011, 2012, 2013
  • Program Committee Member, HCOMP 2012, 2013
  • Reviewer, IUI 2012, 2013
  • Reviewer, CHI 2013
  • Program Committee Member, AAMAS 2011,2012
  • Reviewer, CSCW 2012
  • Senior Program Committee Member, IJCAI 2011
  • Program Committee Member, IUI 2010
  • Program Committee Member, Mixed-Initiative Multi-agent Systems Workshop, AAMAS 2009
  • Reviewer, Transactions on Pattern Analysis and Machine Intelligence
  • Reviewer, Journal of Artificial Intelligence Research

Recent Projects

Debugging Machine Learning with Human Input

As machine learned models being widely used in domains ranging from judiciary to healthcare to autonomous driving, understanding and characterizing their failures in the open world is critical. This project investigates algorithms, models and workflows for the ideal use of human intelligence for debugging of machine learning systems. It aims to address the issues of interpretability, trust and bias while creating a continuous improvement loop for machine learning.

Hybrid Intelligence: Complementing AI Systems with Human Intelligence

Despite recent advances, AI systems are far from being perfect. When these systems are left to function without human assistance, they may occasionally make mistakes or completely fail. The central idea of this project is creating systems that can benefit from human input through their complete life cycle, including training, execution and testing, instead of functioning alone. The project brings ideas from machine learning, planning and optimization to different domains ranging from citizen science to image captioning to arcade games.

Studies of Human Computation

As machine learned models being widely used in domains ranging from judiciary to healthcare to autonomous driving, understanding and characterizing their failures in the open world is critical. This project investigates algorithms, models and workflows for the ideal use of human intelligence for debugging of machine learning systems. It aims to address the issues of interpretability, trust and bias while creating a continuous improvement loop for machine learning.

Formal Models of Teamwork

As machine learned models being widely used in domains ranging from judiciary to healthcare to autonomous driving, understanding and characterizing their failures in the open world is critical. This project investigates algorithms, models and workflows for the ideal use of human intelligence for debugging of machine learning systems. It aims to address the issues of interpretability, trust and bias while creating a continuous improvement loop for machine learning.

Publications

Workshop Papers and Short Papers