Title: Second-order Optimization Made Practical Abstract: Optimization in machine learning, both theoretical and applied, is presently dominated by first-order gradient methods such as stochastic gradient descent.  Higher-order (preconditioned) optimization methods have become far less prevalent, despite compelling theoretical properties, due to their impractical computation, memory and communication costs.  I will present some recent theoretical, algorithmic ... Read more

Open for all biu students interested to learn how to best utilize the available GPU hardware. Wednesday, December 4⋅10:00am – 12:00pm חדר ישיבות 329, הנדסה Please download following materials before session: profilingTutorial Add to calendar: https://calendar.google.com/event?action=TEMPLATE&tmeid=NGNjb2s0azl0OG9qMmdpZjl2Z3Q5cmZzOWQgb3Jlbi5nbGlja21hbkBt&tmsrc=oren.glickman%40gmail.com Post Event Materials: Talk:  https://www.youtube.com/watch?v=m5He_FlhcRo&feature=youtu.be&t=1 Deck:  https://raanan.blob.core.windows.net/biu/A%20Practical%20Guide%20for%20Reducing%20DNN%20Training%20Time%E2%80%8B.pptx?sp=r&st=2019-12-08T12:50:55Z&se=2020-03-18T20:50:55Z&spr=https&sv=2019-02-02&sr=b&sig=6YtI4y5mhfDK3A9zvpuTd3vLMpyNgITRyouHbz9Hhgs%3D Blog Post: https://sagivtech.com/2017/09/19/optimizing-pytorch-training-code/

Title: GANs for Real-Life Video Applications. Abstract: This talk will focus on two research papers that demonstrate the use of generative adversarial networks for real-life video applications, through face and full-body synthesis. The first paper (“Live Face De-Identification in Video”), proposes a novel method for face de-identification, that enables fully automatic video modification at high ... Read more

Speaker: ALON GONEN Date : 19/12/2019 - 12:00 Title : Beyond Worst Case In Machine Learning: The Oracle Model Abstract: In recent years there has been an increasing gap between the success of machine learning algorithms and our ability to explain their success theoretically. Namely, many of the problems that are solved to a satisfactory degree of precision ... Read more

Sanjeev Khudanpur from John-Hopkins University Two Ideas for Improving Automatic Speech Recognition: One Elegant, and One Very Useful! The Kaldi tools for automatic speech recognition (ASR) are being widely used both for research and for large-scale deployments.  Many innovations large and small have gone into keeping Kaldi up-to-date in this fact-moving field.  I will describe how two ... Read more