つい先週，機械翻訳で驚くべき進展がありました． 教師なし機械翻訳がヤバい進化を遂げててびっくりした．たった半年でBLEUスコアを15から25に改善したのブレイクスルーでは？https://t.co/SVQlYYu2Pt 教師なし学習でこのクオリティの機械翻訳できるのまじで感動するし，ちょっと語っていい？ pic.twitter.com/fBllGtTkgb— Ryobot | りょぼっと (@_Ryobot) 2018年4月23日 要約すると教師なし学習でもひと昔前の教師あり学習の機械翻訳に匹敵する性能を獲得できたというのです．この記事では機械翻訳を知らない初心者にもわかるように魔法のような教師なし機械翻訳の仕組みを説明したいと思います． 教師あり学習の限界 機械翻訳はディープラーニングを適用することで急激に進歩した分野の１つだと思います．Google 翻訳はニューラル機械翻訳を導入するこ
本稿ではニューラルネットワーク，誤差逆伝播法，言語モデル，RNN，LSTM，ニューラル機械翻訳の一連の手法について数理的に解説する． 前編の目次 ニューラルネットワーク 順伝播 (Forwardpropagation) 逆伝播 (Backpropagation) リカレントニューラルネットワーク (RNN) Recurrent Neural Network Language Model (RNNLM) Backpropagation Through Time (BPTT) Long Short-Term Memory (LSTM) Gated Recurrent Unit (GRU) RNN のドロップアウトとバッチ正規化 ニューラル機械翻訳 (NMT) Sequence to Sequence (seq2seq) 注意 (Attention) 双方向エンコーダー・多層LSTM 評価手法
NLP2017 NMT Tutorial「ゼロから始めるニューラルネットワーク機械翻訳」
1. The document discusses various statistical and neural network-based models for representing words and modeling semantics, including LSI, PLSI, LDA, word2vec, and neural network language models. 2. These models represent words based on their distributional properties and contexts using techniques like matrix factorization, probabilistic modeling, and neural networks to learn vector representatio
The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experi
Neural Turing Machines ソースコード Attentionインターフェース Adaptive Computation Time コード Neural Programmer ソースコード 総括的な今後の展望 参考 本記事はAttention and Augmented Recurrent Neural Networksの著者の許諾を得て翻訳しました。 Recurrent Neural Networksは、文章や音声、動画などの順序を持つデータをニューラルネットワークで扱うことができるディープラーニングの重要な要素のうちの1つです。 RNNを使うことで、一連の順序に現れるパターンを抽象的に理解して、注釈をつけたり、まったくのゼロから一連のデータを生成することすらできるのです！ シンプルなRNNの設計では、長期の時系列データには苦戦しますが、「long short-term
Google’s Multilingual Neural Machine Translation System: Enabling Zero-Shot Translation Melvin Johnson, Mike Schuster, Quoc V. Le, Maxim Krikun, Yonghui Wu, Zhifeng Chen, Nikhil Thorat melvinp,schuster,qvl,krikun,yonghui,zhifengc,nsthorat@google.com Fernanda Viégas, Martin Wattenberg, Greg Corrado, Macduff Hughes, Jeffrey Dean Abstract We propose a simple, elegant solution to use a single Neural Ma
Language translation is important to Facebook’s mission of making the world more open and connected, enabling everyone to consume posts or videos in their preferred language — all at the highest possible accuracy and speed. Today, the Facebook Artificial Intelligence Research (FAIR) team published research results using a novel convolutional neural network (CNN) approach for language translation t
Since we don’t have any nice NMT (Native Memory Tracking) MBean in HotSpot (yet), and therefore not in the JMC console, I thought I’d show how it can be done using command line arguments and JCMD. Please note that you’ll get a 5-10% performance hit if you enable this. Step 1 – Enabling NMT This is done by using the following command line: -XX:NativeMemoryTracking=[off|summary|detail] Where the dif
Machine translation systems achieve near human-level performance on some languages, yet their effectiveness strongly relies on the availability of large amounts of parallel sentences, which hinders their applicability to the majority of language pairs. This work investigates how to learn to translate when having access to only large monolingual corpora in each language. We propose two model varian
In this paper, we study a new learning paradigm for Neural Machine Translation (NMT). Instead of maximizing the likelihood of the human translation as in previous works, we minimize the distinction between human translation and the translation given by an NMT model. To achieve this goal, inspired by the recent success of generative adversarial networks (GANs), we employ an adversarial training arc
An effective method to improve neural machine translation with monolingual data is to augment the parallel training corpus with back-translations of target language sentences. This work broadens the understanding of back-translation and investigates a number of methods to generate synthetic source sentences. We find that in all but resource poor settings back-translations obtained via sampling or
Philosophy We strive to create an environment conducive to many different types of research across many different time scales and levels of risk. Learn more about our Philosophy Learn more
Current state-of-the-art machine translation systems are based on encoder-decoder architectures, that first encode the input sequence, and then generate an output sequence based on the input encoding. Both are interfaced with an attention mechanism that recombines a fixed encoding of the source tokens based on the decoder state. We propose an alternative approach which instead relies on a single 2
The field of machine translation faces an under-recognized problem because of inconsistency in the reporting of scores from its dominant metric. Although people refer to "the" BLEU score, BLEU is in fact a parameterized metric whose values can vary wildly with changes to these parameters. These parameters are often not reported or are hard to find, and consequently, BLEU scores between papers cann
j次のブックマーク
k前のブックマーク
lあとで読む
eコメント一覧を開く
oページを開く
教師なし学習は機械翻訳に魔法をかけるか？ - ディープラーニングブログ
深層学習による自然言語処理 - RNN, LSTM, ニューラル機械翻訳の理論
Attention Is All You Need
Attentionで拡張されたRecurrent Neural Networks
[PDF]Google’s Multilingual Neural Machine Translation System: Enabling Zero-Shot Translation
A novel approach to neural machine translation
Native Memory Tracking in 7u40 – Marcus Hirt
Phrase-Based & Neural Unsupervised Machine Translation
Adversarial Neural Machine Translation
[1808.09381] Understanding Back-Translation at Scale
Recent Advances in Google Translate
Pervasive Attention: 2D Convolutional Neural Networks for Sequence-to-Sequence Prediction
A Call for Clarity in Reporting BLEU Scores