https://feisky.xyz/machine-learning/rnn.html Recent content in 递归神经网络 on Machine Learning Hugo -- gohugo.io zh-cn https://feisky.xyz/machine-learning/rnn/seq2seq.html Mon, 01 Jan 0001 00:00:00 +0000 https://feisky.xyz/machine-learning/rnn/seq2seq.html Sequence to Sequence来自Google Cho et al., 2014，主要用来解决翻译、智能问答等序列型问题。Seq2Seq由两个不同的RNN（比如GRU或L https://feisky.xyz/machine-learning/rnn/word2vec.html Mon, 01 Jan 0001 00:00:00 +0000 https://feisky.xyz/machine-learning/rnn/word2vec.html 在信号处理领域，图像和音频信号的输入往往是表示成高维度、密集的向量形式，在图像和音频的应用系统中，如何对输入信息进行编码(Encoding) https://feisky.xyz/machine-learning/rnn/sequence.html Mon, 01 Jan 0001 00:00:00 +0000 https://feisky.xyz/machine-learning/rnn/sequence.html tf.SequenceExample tf.SequenceExample可以用来方便的处理序列数据，它包含两部分 context：非序列化的数据 feature_list：序列数 https://feisky.xyz/machine-learning/rnn/01_wikipedia.html Mon, 01 Jan 0001 00:00:00 +0000 https://feisky.xyz/machine-learning/rnn/01_wikipedia.html Wikipedia Word Embeddings To train a set of embedded word vectors, run train.py: $ python3 train.py This will save your word embeddings into ./wikipedia/embeddings.npy Unfortunately, at this time the sample code is only compatible with Python 3. We’re working on providing Python 2 translations of the code. Refer tensorflowbook. https://feisky.xyz/machine-learning/rnn/02_imdb.html Mon, 01 Jan 0001 00:00:00 +0000 https://feisky.xyz/machine-learning/rnn/02_imdb.html IMDB Sentiment Analysis To train the sentiment analysis model, simply run train.py: $ python3 train.py Note that this requires training word embeddings from the previous Wikipedia model. Unfortunately, at this time the sample code is only compatible with Python 3. We’re working on providing Python 2 translations of the code. Refer tensorflowbook. https://feisky.xyz/machine-learning/rnn/03_ocr.html Mon, 01 Jan 0001 00:00:00 +0000 https://feisky.xyz/machine-learning/rnn/03_ocr.html RNN and Bi-directional RNN for OCR To run and train the standard RNN, simply run train.py: $ python3 train.py To run the bi-directional RNN, use train_bidirectional.py: $ python3 train_bidirectional.py Unfortunately, at this time the sample code is only compatible with Python 3. We’re working on providing Python 2 translations of the code. Refer tensorflowbook. https://feisky.xyz/machine-learning/rnn/04_arxiv.html Mon, 01 Jan 0001 00:00:00 +0000 https://feisky.xyz/machine-learning/rnn/04_arxiv.html arXiv character-by-character generative model To train the model, simply run train.py: $ python3 train.py Then, to generate a sample abstract, run sample.py: $ python3 sample.py If you want to change the starting seed of the generated abstract or change its length, just modify this line in sample.py to your liking: print(Sampling(get_params())('<SEED>', <LENGTH>)) Unfortunately, at this time the sample code is only compatible with Python 3. We’re working on providing Python 2 translations of the code.