Natural facial behavior can reveal information about our internal states intentions, so the potential for facial image analysis technology is far-reaching, across fields of healthcare, education, advertising, and retail. Marni Bartlett, Ph.D. is co-Founder and Lead Scientist at Emotient, a San Diego based start-up company for automatic facial expression analysis. Marni is a leader in the field of automatic facial expression analysis, pioneering machine learning approaches to facial expression detection from the nascence of the field in the mid-1990's with Paul Ekman and Terrence Sejnowski.
Marni will be speaking at the Deep Learning Summit later this month, so we caught up with her to hear her views on deep learning and facial analysis technologies.
What are the key factors that have enabled recent advancements in deep learning?
Algorithms for speeding up learning of deep networks have been crucial for recent progress.
What are the main types of problems now being addressed in the Neural Network space?
There is a huge range of problems currently being addressed. One is a resurgence of effort to learn dynamic structure, now that there are more powerful learning tools. Specifically regarding deep nets, there is some interesting work on generalization to new tasks, namely using the middle layers learned for one task as features for another related task. Advantages have also been shown for learning features for multiple related tasks.
What are the practical applications of your work and what sectors are most likely to be affected?
Automatic facial expression analysis has a broad application over many fields. Deep learning provides the robustness to take facial expression out of the lab and into the real world. We can expand beyond applications in market research and advertising, into retail, healthcare, and education. For example, I am presently involved in a project to measure pain in clinical settings. A current issue in hospitals today is that pain is undermanaged, and its particularly a problem in pediatrics. Pain tends to be underestimated by clinical staff. I am collaborating with Rady Children's hospital in San Diego to develop and evaluate a system for measuring the level of pain in clinical settings. The system could provide standardized and continuous patient monitoring that is potentially scalable.
What advancements excite you most in the field?
The improvement in the ability to learn non-linear manifolds from finite amounts of data is a real game changer.
The Deep Learning Summit is taking place in San Francisco on 29-30 January. For more information and to register, go to: re-work.co/deep-learning