Briefing Notes for Skype Call with Green Power Inc. February 29, 2012 Objectives To discover lucrative new problem areas to which transfer learning can be applied. To find early adopters/partners for engaging in transfer learning and development of testimonials. Breif Bio on Daniel Silver, PhD, CIM I am a Professor in and the Director of the Jodrey School of Computer Science at Acadia University. My research focuses on machine learning and its application in data mining, intelligent agents, user modelling, and adaptive systems. I teach related courses as well as courses on software engineering, e-commerce and the impact of computing on our environment and society. I was the President of the Canadian Artificiual Intellgence Association (CAIAC) and currently hold the Past-President position. In 2005, I founded the Acadia University Robot Programming Competitions and have been the FIRST LEGO League (FLL) Partner for Nova Scotia since 2006. In 2011 I accepted Science Champion Award from the Nova Scotia Discovery Center on behalf of all the wonderful people at Acadia and the NSCC that I have had the pleasure to work with on youth robotics and the advancement of STEM education. Since January, 1993, I have operated a consulting business, CogNova Technologies, that offers services in the areas of machine learning, knowledge discovery and data mining. About CogNova Technologies - Founded in 1993, in London, Ontario and is now located in the Annapolis Valley of Nova Scotia - The company's mission is to provide education, consultation, and services in the areas of machine learning, knowledge discovery, and data mining. o Education: Courses and seminars on the theory and application of machine learning and data mining technologies and the knowledge discovery process o Services: Project management,installation and application of third party software, data analysis and model generation using CogNova proprietary systems, and summary and analysis of results. o Consulting: Situation analysis and problem definition, selection of third party systems, project guidance, and trouble shooting. - Prior engagements: o First data mining project for London Life, London, Ontario o First data mining project for MTT, Halifax (now part of Aliant) o Feasibility studies on the use of intelligent adaptive user interfaces by Progeny Software Inc, Wolfville, NS o Training of Federal Government and Nova Scotia Provincial Government employees directly or through Dalhousie University. o Consulting and data mining services to Health Nova Scotia, Halifax. o Data mining services for TSI Associates Development Corporation, PEI. New Offering in Transfer Learning Transfer Learning is an area of machine learning that studies the use of prior task knowledge as a form of inductive bias to improve the effectiveness (more accurate hypotheses) and efficiency (shorter training times) of learning. This is the next generation machine learning technology that 1 uses prior knowledge when learning a complex classification or predictive function (see http://ml3.acadiau.ca/). Danny Silver is recognized as a leading Canadian researcher in this area (see the DBLP list of publications). A great example of the successfully use of transfer learning that we have led is the development of models that can accurately predict the flow rate of streams in the Annapolis Valley from weather conditions one or three days earlier. Normally, 5 to 10 years worth of data is needed to build an accurate model for a stream, however with knowledge transfer from models of nearby streams accurate models can be developed from only 1 year of data for a new stream. See the figure below as well as the slides at the following Share directory. 2