下载  >  人工智能  >  深度学习  > 李宏毅 -生成对抗网络GAN课件 234页

李宏毅 -生成对抗网络GAN课件 234页 评分:

李宏毅 -生成对抗网络GAN课件 234页 李宏毅 -生成对抗网络GAN课件 234页
Yann le cun's comment What are some recent and potentially upcoming breakthroughs in unsupervised learning? Yann Lecun. Director of al research at facebook and professor at nyu Written Jul 29. Upvoted by Joaquin Quinonero Candela, Director Applied Machine Learning at Facebook and Huang Xiao Adversarial training is the coolest thing since sliced bread I've listed a bunch of relevant papers in a previous answer Expect more impressive results with this technique in the coming years What's missing at the moment is a good understanding of it so we can make it work reliably. It's very finicky. Sort of like ConvNet were in the 1990s, when I had the reputation of being the only person who could make them work(which wasnt true) https://www.quora.com/what-are-some-recent-and potentially-upcoming-breakthroughs-in-unsupervised-learning Yann le cun's comment What are some recent and potentially upcoming breakthroughs in deep learning? Yann LeCun. Director of al research at facebook and professor at nyu Written Jul 29. Upvoted by Joaquin Quinionero Candela, Director Applied Machine Learning at Facebook and Nikhil Garg, I lead a team of Quora engineers working on MU/NLP problems The most important one, in my opinion, is adversarial training (also called GaN for Generative Adversarial Networks). This is an idea that was originally proposed by Ian Goodfellow when he was a student with Yoshua Bengio at the University of Montreal (he since moved to Google Brain and recently to OpenAl) This, and the variations that are now being proposed is the nost interesting idea in the last 10 years in ML, in my opinion https://www.quora.com/what-are-soMe-recent-and-potentially-upcoming-breakthroughs in-deep-learning Deep Learning in One Slide( Review Fully connected feedforward network Many kinds of network structures. >Convolutional neural network(CNN) Recurrent neural network rnn) Different networks can take different kinds of input/output Vector Mtex■ Network y Vector seq (speech, video, sentence How to find Given the example inputs/ outputs as the function? training data:{(X1y1),(×2y2),……,(x1 1000y1000 Creation Anime face Generation Drawing? 何之源的知乎:htts:/ zhuanlan. zhihu. com/p/24767059 Dcgan:https://github.com/carpedm20/dcgan-tensorflow Poweredby:http://mattya.github.io/chainer-dcgan/ Basic ldea of gan it is a neural network (NN), or a function vector Generator image matrix 0.1 3 33 Generator Generator 2.4 2.4 0.9 Each dimension of input vector 0.9 Longer hair represents some characteristICs 0.1 0.1 3 Generator Generator 2.4 2.4 0.9 blue hair 3.5 Open mouth Basic ldea of gan it is a neural network (NN), or a function Discri scalar Image minator Larger value means real smaller value means fake Discri Discri 1.0 1.0 minator minator Discri Discri 0.1 minator minator Basic ldea of gan Generator Brown veIns Butterflies are Butterflies do not brown not have veins Discriminator This is where the term adversarialcomes from Basic ldea of gan you can explain the process in different ways NN NN NN Generator Generator Generator V1 V2 V3 Discri Discri Discri minator minator minator V1 V2 V3 Real images

...展开详情
2017-11-30 上传 大小:9.4MB
举报 收藏 (1)
分享

评论 下载该资源后可以进行评论 共4条

u013944699 太贵了,不值得,另外一个人只要三个积分。内容还可以
2018-11-19
回复
fandongwei 234页的PPT,赞,好好学习下
2018-07-27
回复
weixin_38100342 这是我见到的有关GAN讲稿中最好的一个
2017-12-23
回复
weixin_40616987 为什么我每次下载都要下2次,,,
2017-12-07
回复
可能是网络问题,这只能找csdn做技术支持了