Скачать или смотреть AIDA Seminars - Contrastive learning:Intro to supervised&self-supervised approaches - Furkan Yeşiler

Abstract:
Contrastive learning is a learning paradigm where the training process is guided using the similarity relationships between items. This talk aims to give a high-level introduction to supervised and self-supervised contrastive learning approaches. We will start with motivating the use of contrastive methods for discriminative tasks and representation learning. We will then go through the supervised metric learning approaches and discuss the advantages and disadvantages of using pairs, triplets, and proxies. We will continue with introducing several self-supervised approaches for contrastive learning, including systems that use noise contrastive estimation-based loss functions. We will conclude the talk by discussing the recent self-supervised approaches where the need for using a contrastive term is eliminated.

Bio: Furkan Yesiler is an Early Stage Researcher / Ph.D. candidate as a part of the MIP-Frontiers project (MSCA Grant No:765068) at Music Technology Group, Universitat Pompeu Fabra (Barcelona). His Ph.D. research is focused on incorporating deep learning techniques to build accurate and scalable music version identification systems for industrial use-cases. He received his MSc degree in Sound and Music
Computing also at MTG, UPF with his thesis on singing voice research. He graduated summa cum laude with two BSc degrees in computer engineering and industrial engineering from Koc University (Istanbul) where he was accepted with a full scholarship. During his bachelor’s studies, he did internships in management consulting, and mergers and acquisitions advisory companies in Istanbul.

Slides of the talk are accessible from: https://docs.google.com/presentation/...