[cs231n]Assignment #2: Neural Networks, ConvNets I
[cs231n]Assignment #3: ConvNets II, Transfer Learning, Visualization
Assignment Website
Reference
- K. Simonyan, A. Vedaldi, A. Zisserman , “Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps”, ICLR Workshop 2014
- Szegedy, Christian, et al. “Intriguing properties of neural networks.” arXiv preprint, 2013.
- Nguyen, Anh, Jason Yosinski, and Jeff Clune. “Deep Neural Networks are Easily Fooled: High Confidence Predictions for Unrecognizable Images.” arXiv preprint, 2014.
[Caffe](1) Google Protocol Buffer
Caffe代码导读（1）：Protobuf例子
Protocol Buffer Basics
总的来说，Protobuf主要是在不同平台不同语言之间传递变量的作用. 在Caffe中具体的作用就是
一个软件项目 = 数据结构 + 算法 + 参数，对于数据结构和算法我们都已经有较多研究，但不同开发者对参数管理却各有千秋。有人喜欢TXT格式化的参数文件，有人喜欢BIN简单高效，也有人喜欢图形化界面的直观。不一致的参数管理带来很多问题，例如一个项目组内不同成员必须 约定一套统一的参数方案，或者称为通信协议，这样便于模块集成。而Protobuf工具就完美解决了这个问题，关键部分代码自动生成，节省了>> 大量的开发、调试时间
除了上述简单功能，Protobuf还可以用来传递不同语言(C/C++与Java、Python)之间的参数，省去了自己手动维护数据结构的繁琐工作。也可以支持客户端/服务器模式，在主机/从机之间传递参数。
caffe代码中的参数初始化、参数保存操作
Notes for <SEMANTIC IMAGE SEGMENTATION WITH DEEP CONVOLUTIONAL NETS AND FULLY CONNECTED CRFS>
The main difference between this model and other state-of-the-art models is the combination of
pixel-level CRFs and DCNN-based ‘unary terms’.
notes for <Fully Convolutional Networks for Semantic Segmentation>
Use CNN for Semantic Segmentation.(Pixel-level inference)
First trick: Convert the fully connected layer to convolutional layer. The details can be seen in CS231n. (which change the pretrained net to the structure we want. Increase the efficiency.)
Second trick: Get dense predictions from coarse outputs(the coarse outputs are got from the first trick)
- Shift and stich (proposed in Overfeat)
- Upsampling / backwards strided convolution
Use the whole image instead of patch sampling for training?
Combining fine layer and coarse layers to make local predictions that respect global structure.(Get more details)
[cs231n]Working with Caffe, an open-source ConvNet library
[cs231n]Beyond Image Classification
[CS231n]ConvNet Tips and Tricks: squeezing out the last few percent
[cs231n]Transfer Learning and Fine-tuning Convolutional Neural Networks
Lecture Note
Lecture Slides It contains extra material
Reference
- Caffe Model Zoo: Where people share their network.
- Understanding the source of ConvNet performance Visualizing and Understanding Convolutional Networks
- CNN Features off-the-shelf: an Astounding Baseline for Recognition trains SVMs on features from ImageNet-pretrained ConvNet and reports several state of the art results.
- DeCAF reported similar findings in 2013. The framework in this paper (DeCAF) was a Python-based precursor to the C++ Caffe library.
- How transferable are features in deep neural networks? studies the transfer learning performance in detail, including some unintuitive findings about layer co-adaptations.
[cs231n]Understanding and Visualizing Convolutional Neural Networks
Lecture Note
References
Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps
Rich feature hierarchies for accurate object detection and semantic segmentation
Visualizing and Understanding Convolutional Networks
Data Gradient.
Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps
DeconvNet.
Visualizing and Understanding Convolutional Networks
Guided Backpropagation.
Striving for Simplicity: The All Convolutional Net
Reconstructing original images based on CNN Codes
Understanding Deep Image Representations by Inverting Them
How much spatial information is preserved?
Do ConvNets Learn Correspondence? (tldr: yes)
Plotting performance as a function of image attributes
ImageNet Large Scale Visual Recognition Challenge
Fooling ConvNets
Explaining and Harnessing Adversarial Examples
Comparing ConvNets to Human labelers
