Description
量子機械学習で先駆的な研究をされている山崎隼汰氏に、量子情報の基本的な事項から始めて最近の研究内容について解説して頂きます。量子情報と機械学習の新たな交流が生まれることを期待して、ディスカッションの時間を少し長めに設定しています。使用言語は日本語です。
Speaker: Dr. Hayata Yamasaki
Affiliation:
1. Institute for Quantum Optics and Quantum Information (IQOQI), Austrian Academy of Sciences
2. Atominstitut, Technische Universität Wien
Position:Postdoc
Title:
Learning with Optimized Random Features: Quantum Computation for Accelerating Machine Learning
Abstract:
This talk will review the basics of quantum computation, and a series of recent works on quantum machine learning (QML) with optimized random features. The goal of the talk is to explain how to use exponential speedup achieved by quantum computation to accelerate learning without imposing restrictive assumptions.
Random features are a central technique for scalable learning algorithms based on kernel methods. A recent work has shown that an algorithm using quantum computation can exponentially speed up sampling of optimized random features, even without imposing restrictive assumptions on sparsity and low-rankness of matrices that had limited applicability of conventional QML algorithms. This QML algorithm makes it possible to significantly reduce and provably minimize the required number of features for achieving learning tasks.
This talk will present applications of this QML algorithm to significant acceleration of leading regression and classification algorithms based on kernel methods, based on the following papers.
https://arxiv.org/abs/2004.10756
https://arxiv.org/abs/2106.09028
