Stéphane Mallat: "Deep Generative Networks as Inverse Problems"
New Deep Learning Techniques 2018
"Deep Generative Networks as Inverse Problems"
Stéphane Mallat, École Normale Supérieure
Abstract: Generative Adversarial Networks and Variational Auto-Encoders provide impressive image generations from Gaussian white noise, which are not well understood. We show that such generations do not require to learn a discriminator or an encoder. They are computed with a scattering transform which preserve the deformation properties of image synthesis. The deep convolutional network generator is calculated as the solution of a regularized inverse problem. We show that this approach also applies to time-series and audio synthesis, thus providing an alternative to recurrent neural networks and wavenets. Numerical results will be shown on images and audio signals.
Joint work with Tomas Angles and Mathieu Andreux, École Normale Supérieure, Collège de France.
Institute for Pure and Applied Mathematics, UCLA
February 6, 2018
For more information: http://www.ipam.ucla.edu/programs/workshops/new-deep-learning-techniques/?tab=overview
Видео Stéphane Mallat: "Deep Generative Networks as Inverse Problems" канала Institute for Pure & Applied Mathematics (IPAM)
"Deep Generative Networks as Inverse Problems"
Stéphane Mallat, École Normale Supérieure
Abstract: Generative Adversarial Networks and Variational Auto-Encoders provide impressive image generations from Gaussian white noise, which are not well understood. We show that such generations do not require to learn a discriminator or an encoder. They are computed with a scattering transform which preserve the deformation properties of image synthesis. The deep convolutional network generator is calculated as the solution of a regularized inverse problem. We show that this approach also applies to time-series and audio synthesis, thus providing an alternative to recurrent neural networks and wavenets. Numerical results will be shown on images and audio signals.
Joint work with Tomas Angles and Mathieu Andreux, École Normale Supérieure, Collège de France.
Institute for Pure and Applied Mathematics, UCLA
February 6, 2018
For more information: http://www.ipam.ucla.edu/programs/workshops/new-deep-learning-techniques/?tab=overview
Видео Stéphane Mallat: "Deep Generative Networks as Inverse Problems" канала Institute for Pure & Applied Mathematics (IPAM)
Показать
Комментарии отсутствуют
Информация о видео
17 февраля 2018 г. 3:34:44
00:37:10
Другие видео канала
Professor Stéphane Mallat: "High-Dimensional Learning and Deep Neural Networks"Inverse Problems Lecture 7/2017: computational model for 2D tomography 1/5Learning to Solve Inverse Problems in Imaging - Willet - Workshop 1 - CEB T1 2019Accelerating ptychography with spectral initializationsFlorent Krzakala | Generative models are the new sparsityStéphane Mallat - Learning Physics with Deep Neural Networks (October 17, 2018)Data-driven regularisation for solving inverse problems - Carola-Bibiane Schönlieb, Turing/CambridgeDeep Image Prior | Two Minute Papers #219Stéphane Mallat: "Scattering Invariant Deep Networks for Classification, Pt. 3"ECMI2021 Talk Carola-Bibiane Schönlieb, "Deep Learning for Solving Inverse Imaging Problems"[CVPR 2020] Self-Supervised Monocular Scene Flow Estimation (5-min talk)MIA: Josh Batson, Noise2Self: Blind denoising by self-supervisionAge Prediction From Facial Images (CNN Regression) - Data Every Day #200Stéphane Mallat: Modeling Deep Networks: Network Learning for Image ProcessingThe mother of all representer theorems for inverse problems & machine learning - Michael UnserUllrich Köthe | Analyzing Inverse Problems in Natural Science using Invertible Neural NetworksRebecca Willett: "Learning to Solve Inverse Problems in Imaging"Understanding Wavelets, Part 5: Machine Learning and Deep Learning with Wavelet ScatteringStéphane Mallat: A Wavelet Zoom to Analyze a Multiscale World