##Deep Network Workbench

###Caffe

* [Caffe: a fast open framework for deep learning](https://github.com/BVLC/caffe)
* [DIY Deep Learning for Vision: a Hands-On Tutorial with Caffe](https://github.com/BVLC/caffe-tutorial/blob/master/index.md)
* [Caffe Model Zoo](https://github.com/BVLC/caffe/wiki/Model-Zoo)
* [DeepNetworks by C++](http://blog.csdn.net/linger2012liu/article/category/2146595) //我所写的CNN框架 VS caffe - linger
* [Caffe Hands on Notes](https://yufeigan.github.io/)
* [深度学习caffe的代码怎么读？](http://www.zhihu.com/question/27982282)
* 

###Torch, etc

* [Torch](http://torch.ch/)

##DQN

* https://github.com/deepmind

##@Nvidia.Jetson

* [Nvidia.Jetson](https://developer.nvidia.com/get-started-jetson)
* [Linux For Tegra Archive](https://developer.nvidia.com/embedded/linux-tegra)
* [How to run the Caffe deep learning vision library on Nvidia’s Jetson mobile GPU board](http://petewarden.com/2014/10/25/how-to-run-the-caffe-deep-learning-vision-library-on-nvidias-jetson-mobile-gpu-board/)

##the World's First Makerable Deep Learning Robot

integrated with:

* Nvidia [Jetson](https://developer.nvidia.com/get-started-jetson) Embedded Tegra Computing
* Autonomous Vehicle
* Caffe
* DQN

##Neuromorphic Processing: A New Frontier in Scaling Computer Architecture

* [qualcomm zeroth](https://www.qualcomm.com/invention/cognitive-technologies/zeroth)
 * https://www.qualcomm.com/invention/cognitive-technologies
* [体系结构研究者眼中的神经网络硬件](http://mp.weixin.qq.com/s?__biz=MzA4MjE5NjAzMg==&mid=208687334&idx=1&sn=b6f6cfd24d484c835738a6cbb8376f76&scene=1&from=singlemessage&isappinstalled=0#rd)
* [计算所的DianNao和DaDianNao为什么能连续斩获ASPLOS'14和MICRO'14的Best Paper？](http://www.zhihu.com/question/29269842/answer/55912804)
* CM1K
 * http://techcrunch.com/2015/01/31/the-ongoing-quest-for-the-brain-chip/
 * https://www.indiegogo.com/projects/braincard-pattern-recognition-for-all#/story
 * http://neuromorthings.com/#about

###Open Problems

Several problems to address

* 以下问题相互递进
 * 如何主动发现问题解决问题
 * 如何使用先验知识，将先验知识转移进网络中
 * 已训练的网络训练新知的过程怎样变迁
* 深度神经网络的数据通量
 * 数据接口带宽，存储量
 * 
* 深度硬件，运动控制，时间序列，小脑神经网络，Sequential Deep Learning，Deep time series
- Currently state of the art but...([Scyfer, Universiteit van Amsterdam spin off](http://scyfer.nl/wp-content/uploads/2014/06/Neural-Networks-and-Deep-Learning-1.pdf))
 - No way of doing logical inference (extrapolation)
 - No easy integration of abstract knowledge
 - Hypothetic space bias might not conform with realityWhen to apply Deep
* 看上去一切触手可及对不对？可是我们离真正的人工智能还有多远呢？至少还需要有两三个这样的变革吧 [zhihu](http://www.zhihu.com/question/31430100/answer/51997378)
 
##News

* [New ‘deep learning’ technique enables robot mastery of skills via trial and error](http://news.berkeley.edu/2015/05/21/deep-learning-robot-masters-skills-via-trial-and-error/), UC Berkeley