本文为美国麻省理工学院（作者：Chiyuan Zhang）的博士论文，共150页。

近年来，深度学习在视觉对象识别、检测与分割、自动语音识别、自然语言处理和强化学习等领域取得了成功的应用。在这篇论文中，我们将从不同的角度来研究深度学习。

首先，我们将研究泛化问题，这是机器学习理论中最基本的概念之一。我们将展示在深度学习的情况下，泛化的特征如何与传统的方法不同，并提出解决方法。

从理论到实践，我们将展示两种不同的深度学习应用。一个是源于从地震记录中自动检测地球物理特征以帮助油气勘探的现实世界问题；另一个是源于对人类听觉系统的计算机神经科学建模和研究。更具体地说，我们将展示深度学习如何适应与来自不同领域的问题相关联的独特结构。

最后，我们转向计算机系统设计的角度，展示我们在构建更好的深度学习系统方面所做的努力，以便在学术界和工业界实现高效和灵活的计算。

In the recent years deep learning has witnessed successful applications in many different domains such as visual object recognition, detection and segmentation, automatic speech recognition, natural language processing, and reinforcement learning. In this thesis, we will investigatedeep learning from a spectrum of different perspectives.

First of all, we will study the question of generalization, which is one of the most fundamental notion in machine learning theory. We will show how, in the regime of deep learning, thecharacterization of generalization becomes different from the conventional way, and proposealternative ways to approach it.

Moving from theory to more practical perspectives, we will show two different applicationsof deep learning. One is originated from a real world problem of automatic geophysical featuredetection from seismic recordings to help oil & gas exploration; the other is motivated from acomputational neuroscientific modeling and studying of human auditory system. More specifically, we will show how deep learning could be adapted to play nicely with the unique structuresassociated with the problems from different domains.

Lastly, we move to the computer system design perspective, and present our efforts in building better deep learning systems to allow efficient and flexible computation in both academicand industrial worlds.

1.  引言

2.  深度学习中的泛化理解

3.  具有域结构深度学习的应用

4.  构建灵活与高效的深度学习系统

5.  结论

更多精彩文章请关注公众号：