Welcome!

The NIPS-10 Workshop on
Machine Learning for Assistive Technologies

To be held at the Twenty-fourth Annual Conference on Neural Information Processing Systems (NIPS-10)
December 10, 2010 in Whistler, British Columbia, Canada

Workshop Schedule Important Dates Submissions Due October 20, 2010 Notification of Acceptance November 3, 2010 Workshop Date December 10, 2010 Workshop Information Participants	Workshop Goals This workshop will expose the research area of assistive technology to machine learning specialists, will provide a forum for machine learning researchers and medical/industrial practitioners to brainstorm about the main challenges, and will lead to developments of new research ideas and directions in which machine learning approaches are applied to complex assistive technology problems. The workshop will discuss important open questions aimed at the next five years of research in a number of key areas. More details follow below. Confirmed Invited Speakers Prasad Tadepalli: A Decision-Theoretic Framework for Assistive Technologies Matthai Philipose: Building Machines for Care Call for Contributions The organizing committee is currently seeking either technical papers (eight pages in the conference format) or else abstracts (up to two pages) describing research relevant to the workshop. Submissions should be sent via email to mlat.nips2010@gmail.com and should be in Postscript, PDF, or MS Word format. Please adhere to the NIPS style (see http://nips.cc/PaperInformation/StyleFiles) Previously published work that is reworded, summarized or extended may be submitted to the workshop. However, priority will be given to novel work. Papers do not need to be blinded and will be reviewed by the organising committee for suitability in the workshop. Accepted papers will be presented as posters. Exceptional work will be considered for oral presentation. Papers will be collected and distributed as workshop notes (non-archival) at the conference. If the papers are of sufficient quantity and quality, we will seek to publish them as an edited book or journal special issue. Please contact the organisers (see below) if you would like to submit or attend. Workshop Format Participants will be machine learning specialists with an interest in expanding their research profile into the area of assistive technology, existing researchers in AT, practitioners in occupational therapy with an interest in machine learning, and technology developers with an interest in further developing their application area into this novel field of research. The main focus of the workshop will be on discussions and brainstorming sessions of breakout groups with the explicit goal of identifying demands from the field of AT, and ML related research topics that will help to overcome current bottlenecks for successful AT approaches. The workshop will consist of invited talks from two perspectives (medical/industrial and academic/research) to be given by experts from the field. Participants of the workshop will be asked to submit short or long papers. Accepted papers will briefly be presented orally in short (spotlight) sessions. Accompanying posters will be displayed throughout the whole workshop. The workshop will then define breakout discussion topics, and will allocate participants to groups for brainstorming sessions, closing with presentations and discussions. Significant time will be allocated to these breakout discussions and the presentations of their findings.

Overview: An aging demographic has been identified as a challenge for healthcare provision, with technology tipped to play an increasingly significant role. Already, assistive technologies for cognitive and physical disabilities are being developed at an increasingly rapid rate. However, the use of complex technological solutions by specific and diverse user groups is a significant challenge for universal design. For example, 'smart homes' that recognise inhabitant activities for assessment and assistance have not seen significant uptake by target user groups. The reason for this is primarily that user requirements for this type of technology are very diverse, making a single universal design extremely challenging. Adaptivity, the automatic tailoring of solutions for diverse and changing user needs, is therefore a key requirement for the deployment and uptake of complex assistive technology. For example, persons with Alzheimer's disease or related dementias present a large variety of functional difficulties, and each individual may require tailored solutions that are adaptive over time for all but the simplest types of assistive technologies. Machine learning techniques are therefore playing an increasing role in allowing assistive technologies to be adaptive to persons with diverse needs. However, the ability to adapt to these needs carries a number of theoretical challenges and research directions, including but not limited to decision making under uncertainty, sequence modeling, activity recognition, active learning, hierarchical models, sensor networks, computer vision, preference elicitation, interface design and game theory. This workshop will expose the research area of assistive technology to machine learning specialists, will provide a forum for machine learning researchers and medical/industrial practitioners to brainstorm about the main challenges, and will lead to developments of new research ideas and directions in which machine learning approaches are applied to complex assistive technology problems. The workshop will discuss important open questions aimed at the next five years of research in a number of key areas, for example What are the main bottlenecks that are currently holding back complex assistive technologies from being widely deployed/used? The argument to be presented and discussed at the workshop is that the application of adaptivity and machine learning is one of these bottlenecks. However, other viewpoints will be presented and discussed. For example, there may be a lack of economic and social incentives for the uptake of technology. In many types of assistive technologies, the value to the end customer is not always clear, and people are unwilling to pay a high price for some sophisticated technology unless it impacts their life in a significant way. Similarly, in many countries, payment is only issued when there is a human consultation, so healthcare workers have no incentive to use technologies that reduce the need for human involvement. Another example is the scalability of sensor networks and the lack of sufficiently detailed data. Brute-force approaches that rely on massive datasets have been attempted: have they been successfull? Do assistive technologies need some new type of machine learning? Are there any new machine learning problems or is it mostly a matter of adapting existing machine learning techniques to assistive technologies? A key challenge for assistive technologies is the detection of novel or changing patterns of behavior. Are existing novelty detection, feature selection and unsupervised learning techniques sufficient to handle this challenge? Another key challenge is the integration of human behavior modeling and learning with preference learning, adaptability by end users, and decision making. The workshop will expose these interstitial areas and examine what new machine learning challenges exist there. What are the bottlenecks for the scaling of machine learning techniques for the assistive technology domain? More precisely, how can ML algorithms scale to large domains both in terms of state, action and observation spaces, and in terms of temporal extent? Unsupervised learning, feature selection, distributivity, and hierarchy are obvious choices. However, user adaptability and customizability, the appropriate integration of prior knowledge, and the rapid and inexpensive deployment of large sensor networks (including cameras) also play a significant role.

Workshop Organizing Committe

Jesse Hoey School of Computer Science University of Waterloo Email: jhoey@cs.uwaterloo.ca WWW: http://www.cs.uwaterloo.ca/~jhoey	Pascal Poupart School of Computer Science University of Waterloo Email: ppoupart@cs.uwaterloo.ca WWW: http://www.cs.uwaterloo.ca/~ppoupart
Thomas Ploetz School of Computing Science Newcastle University Email: t.ploetz@ncl.ac.uk

Last changed Friday, October 1, 2010