香港科技大学发布的极简 TensorFlow 入门教程。第一天先介绍了深度学习和机器学习的潜力与基本概念。第二天详细地讨论了卷积神经网。第三天详解了循环神经网络。

1. 数据加解密及密态计算，不同数据的计算互不影响 2. 算法逻辑简单，但重复执行次数巨大 (重复的轻量级 3. 数据以批量形式产生，并且数据量巨大 (批量大数

HKUST, Machine Learning notes, D.Y. Yeung。 香港科技大学，机器学习，研究生经典课件。

结合图文及TensorFlow代码，系统介绍线性回归、逻辑回归、SoftMax分类器、神经网络，卷积神经网络CNN，Dropout、优化器、图像识别、MNIST手写识别、GoogLeNet\AlexNet\ResNet\AlphaGo

P2P network is factually an overlay network for distributed object store, search and sharing. This paper will present a survey about it’s recent. First, we will introduce a simple definition of P2P, and define some common operation procedures in P2P. Second, P2P performance metric will be discussed in order to understand and differentiated the practical P2P architectures and protocols. Third, the current P2P architecture will be classified and compared in detail. Then, object search protocols will be carefully discussed with corresponding P2P architecture. We also review the current P2P modeling outputs and list the general modeling approaches. Finally, several novel applications based on P2P techniques and future research directions of P2P overlay networks are briefly investigated. This paper aims to give a clear and complete survey of P2P networks according to clues of its development and evolution process.

本文介绍34页小样本学习综述《Generalizing from a Few Examples: A Survey on Few-Shot Learning》，包含166篇参考文献，来自第四范式和香港科技大学习的研究学者。

香港科技大学Tensorflow三天速成课件（3）可帮助新手快速入门人工智能与机器学习领域，并帮助其熟悉Google 的Tensorflow 的使用