Recent Advances in Unsupervised Deep Learning
Abstract: Most of deep learning algorithms are supervised, requiring a large amount of paired input-output data to train the parameters (e.g. DNN weights) in the learning systems. Such data are often very expensive to acquire in many practical applications. Unsupervised learning is aimed to eliminate the use of such costly training data in learning the system parameters, and is expected to become a new driving force for the future breakthroughs in artificial intelligence (AI) applications.
The key to successful unsupervised learning is to intelligently exploit rich sources of world knowledge and prior information, including inherent statistical structures of input and output, nonlinear (bi-directional) relations between input and output (both inside and outside the application domains), and distributional properties of input/output sequences. In this keynote, I will present a set of recent experiments on unsupervised learning in sequential classification tasks. The novel unsupervised learning algorithm to be described, inspired by concepts from cryptography research, carefully explores the statistical structure in output sequences, and is shown to achieve classification accuracy comparable to the fully supervised system.
Li Deng recently joined Citadel, one of the most successful investment firms in the world, as its Chief AI Officer. Prior to Citadel, he was Chief Scientist of AI and Partner Research Manager at Microsoft. Prior to Microsoft, he was a tenured Full Professor at the University of Waterloo in Ontario, Canada as well as teaching/research at MIT (Cambridge), ATR (Kyoto, Japan) and HKUST (Hong Kong). He is a Fellow of the IEEE, a Fellow of the Acoustical Society of America, and a Fellow of the ISCA. He has been Affiliate Professor at University of Washington (since 2000).
He was on Board Governors of the IEEE Signal Processing Society and of Asian-Pacific Signal and Information Processing Association (2008-2010). He was Editors-in-Chief of IEEE Signal Processing Magazine (2009-2011) and IEEE/ACM Transactions on Audio, Speech, and Language Processing (2012-2014), for which he received the IEEE SPS Meritorious Service Award. He received the 2015 IEEE SPS Technical Achievement Award for his 'Outstanding Contributions to Automatic Speech Recognition and Deep Learning' and numerous best paper and scientific awards for the contributions to artificial intelligence, machine learning, multimedia signal processing, speech and human language technology, and their industrial applications.
Talking to The Brain in Its Own Language
Abstract: Neuroscience research has focused largely on listening to the brain - on taking recordings, analyzing responses, and trying to extract meaning from them. But now we’re entering a new phase where we can go beyond listening and can start talking back to the brain, and we have developed the ability to do it in the brain’s own language. This opens the door to new technologies for treating disease. Here we present the development of one such technology: a new kind of neuro-prosthetic for treating blindness. Implications for machine vision will also be presented.
Sheila Nirenberg is a neuroscientist/professor at Weill Medical College of Cornell University, where she studies neural coding – that is, how the brain takes information from the outside world and encodes it in patterns of electrical activity. The idea is to be able to decode the activity, to look at a pattern of electrical pulses and know what an animal is seeing or thinking or feeling. Recently, she’s been using this work to develop new kinds of prosthetic devices, particularly ones for treating blindness.
Technologies and Sensors for Next Generation Underwater Exploration
James A.R. McFarlane
Abstract: Research in the world’s oceans is no longer a matter of curiosity, it is imperative to the future understanding of the planet we call home. Humankind has imparted unfathomable pressure on what is the foundation of the existence of life on earth. Over fishing, ocean dumping, pollution, ocean acidification, Inter Alia have all contributed to the extreme pressures our oceans are facing. The use of diverse technology to research this critical component of our existence is of paramount importance. Development of advanced sensing or visualization sub-systems are maturing and evolving. These new sensors and samplers are providing us with insights into the ocean realms that were not possible only a few years ago. The development of new tools and technologies must continue just as the research into the depths of our oceans needs to expand. This is critical to the survival of the planet and we will explore the utilization of different vehicle types to accomplish diverse research requirements.
James A.R. McFarlane joined International Submarine Engineering Ltd. (ISE) in September of 2011 as Executive Vice President.
His early career was spent manufacturing underwater vehicles at ISE in Port Coquitlam, British Columbia, Canada. He then worked at sea in the offshore oil and gas industry in Canada, Norway, Scotland and the United States operating Remotely Operated Vehicles (ROV) for various international commercial diving companies. In 1987 he joined Monterey Bay Aquarium Research Institute (MBARI) as Chief ROV Pilot and departed in 2002 as one of the Institutes Directors. He was Vice President at Sound Ocean Systems in Redmond, Washington from 2007 until 2009 while maintaining and operating his marine consulting company that he started in 2002. In September 2009 he was appointed as the Head of the Office of Resources and Environmental Monitoring for the International Seabed Authority. In this capacity, he was responsible for the management of all mineral resources in the deep ocean located in the Area as defined in Part XI of the United Nations Convention on the Law of the Sea.
Mr. McFarlane’s career has included systems development, manufacturing, marine operations, training and management for a broad range of customers. These efforts were for academia, offshore oil and gas, military and educational programs. He was seconded from MBARI to train and lead Canadian Navy forces for wreckage recovery from the Swissair 111 disaster off the coast of Nova Scotia in 1998. His experience in all facets of subsea development, research and operations have provided him with a unique perspective on oceanographic endeavors and the marine environment. Mr. McFarlane is an experienced ROV pilot and has logged over 10,000 hours of piloting time, primarily for oceanographic research.
Mr. McFarlane was the co-chairperson of the National Science Foundation, National Visiting Committee, for the Marine Advanced Technology Center (MATE) for education from 1999 to 2009. He has also been directly involved in numerous film, television and documentary programs highlighting oceanographic research, education and current news. He is active in the scientific oceanographic community and is an active speaker at workshops and conferences internationally. He is a Member of the Marine Technology Society and a Fellow National (’97) of the Explorers Club.