嵌入式AI 双周简报 (2017-07-24)

业界新闻

• OpenBLAS发布0.2.20版本

• [Clarifai launches SDK for training AI on your iPhone

  VentureBeat](https://venturebeat.com/2017/07/12/clarifai-launches-sdk-for-running-ai-on-your-iphone/)

• [We ported CAFFE to HIP - and here’s what happened…

  GPUOpen](http://gpuopen.com/ported-caffe-hip-heres-happened/)

论文

• [1707.01083] ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices
• [1608.06993] Densely Connected Convolutional Networks
• [1707.01209] Model compression as constrained optimization, with application to neural nets. Part I: general framework
• [1707.04319] Model compression as constrained optimization, with application to neural nets. Part II: quantization
• [1707.03718] LinkNet: Exploiting Encoder Representations for Efficient Semantic Segmentation
• [1707.06342] ThiNet: A Filter Level Pruning Method for Deep Neural Network Compression
• [1707.04693] Binarized Convolutional Neural Networks with Separable Filters for Efficient Hardware Acceleration

开源项目

• Caffe2-iOS demo
• Tencent/ncnn: ncnn is a high-performance neural network inference framework optimized for the mobile platform
• jiaxiang-wu/quantized-cnn: An efficient framework for convolutional neural networks
• KimDarren/FaceCropper: Crop faces, inside of your image, with iOS 11 Vision api.
• csarron/emdl: Embedded and mobile deep learning research resources

博文

• 基于OpenGL ES 的深度学习框架编写 - jxt1234and2010的专栏
• 深度学习技术的应用和思考-博客-云栖社区-阿里云

硬件加速

• [解密又一个xPU：Graphcore的IPU

  唐杉 StarryHeavensAbove ](https://mp.weixin.qq.com/s?__biz=MzI3MDQ2MjA3OA==&mid=2247484282&idx=1&sn=65db58a2610e3828c2feea0c6e3de624&chksm=ead1fe6bdda6777dee462d197e7e6d257d312a1879faab13dd2907f507bbe79a3b314b848ac4&mpshare=1&scene=1&srcid=0724s2sVE9TRaszKKRCqSEcB#rd)

Editor: 张先轶、袁帅

本作品采用知识共享署名-相同方式共享 2.0 通用许可协议进行许可。