Abstract:
Deep learning is the force behind some of the recent notable technological advancements and also fake news around AI. Although useful it can be expensive computationally. This in turns limits its usefulness when there not enough compute resources (IoT) or when it is simply too expensive to use (cloud is not free). In this talk, we review what deep learning is from a point of view which facilitates the rest of the discussion. Then we look at some practical ways of shrinking model sizes. We particularly are interested in methods that can result in a reduction on the order of 4x-100x.
Speaker : Alireza Kenarsari
Alireza is the founder of Picovoice (https://picovoice.ai/) - a Canadian AI startup building voice-enabled AI technology that can be embedded into any product. Prior to Picovoice Alireza was a Senior Engineer at Amazon and has been also an early-stage engineer in a few successful technology startups. He is the inventor of five US patents within the fields of deep learning and speech recognition.