Jump to Content
Oriol Vinyals

Oriol Vinyals

Oriol Vinyals is a Principal Scientist at Google DeepMind, and a team lead of the Deep Learning group. His work focuses on Deep Learning and Artificial Intelligence. Prior to joining DeepMind, Oriol was part of the Google Brain team. He holds a Ph.D. in EECS from the University of California, Berkeley and is a recipient of the 2016 MIT TR35 innovator award. His research has been featured multiple times at the New York Times, Financial Times, WIRED, BBC, etc., and his articles have been cited over 70000 times. His academic involvement includes program chair for the International Conference on Learning Representations (ICLR) of 2017, and 2018. He has also been an area chair for many editions of the NeurIPS and ICML conferences. Some of his contributions such as seq2seq, knowledge distillation, or TensorFlow are used in Google Translate, Text-To-Speech, and Speech recognition, serving billions of queries every day, and he was the lead researcher of the AlphaStar project, creating an agent that defeated a top professional at the game of StarCraft, achieving Grandmaster level, also featured as the cover of Nature. At DeepMind he continues working on his areas of interest, which include artificial intelligence, with particular emphasis on machine learning, deep learning and reinforcement learning.
Authored Publications
Google Publications
Other Publications
Sort By
  • Title
  • Title, descending
  • Year
  • Year, descending
    Emergent abilities of large language models
    Barret Zoph
    Colin Raffel
    Dani Yogatama
    Jason Wei
    Liam B. Fedus
    Maarten Paul Bosma
    Percy Liang
    Sebastian Borgeaud
    Tatsunori B. Hashimoto
    Yi Tay
    TMLR (2022)
    Preview abstract Scaling up language models has been shown to predictably confer a range of benefits such as improved performance and sample efficiency. This paper discusses an unpredictable phenomenon that we call emergent abilities of large language models. Such emergent abilities have close to random performance until evaluated on a model of sufficiently large scale, and hence their emergence cannot be predicted by extrapolating a scaling law based on small-scale models. The emergence of such abilities suggests that additional scaling could further expand the range of tasks that language models can perform. We discuss the implications of these phenomena and suggest directions for future research. View details
    Pointer Graph Networks
    Matthew C. Overlan
    Razvan Pascanu
    Charles Blundell
    Thirty-fourth Conference on Neural Information Processing Systems (NeurIPS 2020) (2020) (to appear)
    Preview abstract Graph neural networks (GNNs) are typically applied to static graphs that are assumed to be known upfront. This static input structure is often informed purely by insight of the machine learning practitioner, and might not be optimal for the actual task the GNN is solving. In absence of reliable domain expertise, one might resort to inferring the latent graph structure, which is often difficult due to the vast search space of possible graphs. Here we introduce Pointer Graph Networks (PGNs) which augment sets or graphs with additional inferred edges for improved model expressivity. PGNs allow each node to dynamically point to another node, followed by message passing over these pointers. The sparsity of this adaptable graph structure makes learning tractable while still being sufficiently expressive to simulate complex algorithms. Critically, the pointing mechanism is directly supervised to model long-term sequences of operations on classical data structures, incorporating useful structural inductive biases from theoretical computer science. Qualitatively, we demonstrate that PGNs can learn parallelisable variants of pointer-based data structures, namely disjoint set unions and link/cut trees. PGNs generalise out-of-distribution to 5x larger test inputs on dynamic graph connectivity tasks, outperforming unrestricted GNNs and Deep Sets. View details
    Reinforced Genetic Algorithm Learning for Optimizing Computation Graphs
    Aditya Paliwal
    Felix Gimeno
    Vinod Gopal Nair
    Yujia Li
    Miles Lubin
    International Conference on Learning Representations (ICLR) (2020)
    Preview abstract We present a deep reinforcement learning approach to minimizing the execution cost of neural network computation graphs in an optimizing compiler. Unlike earlier learning-based works that require training the optimizer on the same graph to be optimized, we propose a learning approach that trains an optimizer offline and then generalizes to previously unseen graphs without further training. This allows our approach to produce high-quality execution decisions on real-world TensorFlow graphs in seconds instead of hours. We consider two optimization tasks for computation graphs: minimizing running time and peak memory usage. In comparison to an extensive set of baselines, our approach achieves significant improvements over classical and other learning-based methods on these two tasks. View details
    Preview abstract Due to the phenomenon of "posterior collapse," current latent variable generative models pose a challenging design choice that either weakens the capacity of the decoder or requires augmenting the objective so it does not only maximize the likelihood of the data. In this paper, we propose an alternative that utilizes the most powerful generative models as decoders, whilst optimising the variational lower bound all while ensuring that the latent variables preserve and encode useful information. Our proposed δ-VAEs achieve this by constraining the variational family for the posterior to have a minimum distance to the prior. For sequential latent variable models, our approach resembles the classic representation learning approach of slow feature analysis. We demonstrate the efficacy of our approach at modeling text on LM1B and modeling images: learning representations, improving sample quality, and achieving state of the art log-likelihood on CIFAR-10 and ImageNet 32×32. View details
    Universal Transformers
    Stephan Gouws
    Jakob Uszkoreit
    Lukasz Kaiser
    ICLR (2019)
    Preview abstract Recurrent neural networks (RNNs) sequentially process data by updating their state with each new data point, and have long been the de facto choice for sequence modeling tasks. However, their inherently sequential computation makes them slow to train. Feed-forward and convolutional architectures have recently been shown to achieve superior results on some sequence modeling tasks such as machine translation, with the added advantage that they concurrently process all inputs in the sequence, leading to easy parallelization and faster training times. Despite these successes, however, popular feed-forward sequence models like the Transformer fail to generalize in many simple tasks that recurrent models handle with ease, e.g. copying strings or even simple logical inference when the string or formula lengths exceed those observed at training time. We propose the Universal Transformer (UT), a parallel-in-time self-attentive recurrent sequence model which can be cast as a generalization of the Transformer model and which addresses these issues. UTs combine the parallelizability and global receptive field of feed-forward sequence models like the Transformer with the recurrent inductive bias of RNNs. We also add a dynamic per-position halting mechanism and find that it improves accuracy on several tasks. In contrast to the standard Transformer, under certain assumptions UTs can be shown to be Turing-complete. Our experiments show that UTs outperform standard Transformers on a wide range of algorithmic and language understanding tasks, including the challenging LAMBADA language modeling task where UTs achieve a new state of the art, and machine translation where UTs achieve a 0.9 BLEU improvement over Transformers on the WMT14 En-De dataset. View details
    Preview abstract Voice-activity-detection (VAD) is the task of predicting where in the utterance is speech versus background noise. It is an important first step to determine when to open the microphone (i.e., start-of- speech) and close the microphone (i.e., end-of-speech) for streaming speech recognition applications such as Voice Search. Long short- term memory neural networks (LSTMs) have been a popular archi- tecture for sequential modeling for acoustic signals, and have been successfully used for many VAD applications. However, it has been observed that LSTMs suffer from state saturation problems when the utterance is long (i.e., for voice dictation tasks), and thus requires the LSTM state to be periodically reset. In this paper, we propse an alter- native architecture that does not suffer from saturation problems by modeling temporal variations through a stateless dilated convolution neural network (CNN). The proposed architecture differs from con- ventional CNNs in three respects (1) dilated causal convolution, (2) gated activations and (3) residual connections. Results on a Google Voice Typing task shows that the proposed architecture achieves 14% rela- tive FA improvement at a FR of 1% over state-of-the-art LSTMs for VAD task. We also include detailed experiments investigating the factors that distinguish the proposed architecture from conventional convolution. View details
    A Study on Overfitting in Deep Reinforcement Learning
    Chiyuan Zhang
    Remi Munos
    Samy Bengio
    arXiv (2018)
    Preview abstract Recent years have witnessed significant progresses in deep Reinforcement Learning (RL). Empowered with large scale neural networks, carefully designed architectures, novel training algorithms and massively parallel computing devices, researchers are able to attack many challenging RL problems. However, in machine learning, more training power comes with a potential risk of more overfitting. As deep RL techniques are being applied to critical problems such as healthcare and finance, it is important to understand the generalization behaviors of the trained agents. In this paper, we conduct a systematic study of standard RL agents and find that they could overfit in various ways. Moreover, overfitting could happen "robustly": commonly used techniques in RL that add stochasticity do not necessarily prevent or detect overfitting. In particular, the same agents and learning algorithms could have drastically different test performance, even when all of them achieve optimal rewards during training. The observations call for more principled and careful evaluation protocols in RL. We conclude with a general discussion on overfitting in RL and a study of the generalization behaviors from the perspective of inductive bias. View details
    Hierarchical Representations for Efficient Architecture Search
    Hanxiao Liu
    Karen Simonyan
    Chrisantha Fernando
    Koray Kavukcuoglu
    International Conference on Learning Representations (2018)
    Preview abstract We explore efficient neural architecture search methods and show that a simple yet powerful evolutionary algorithm can discover new architectures with excellent performance. Our approach combines a novel hierarchical genetic representation scheme that imitates the modularized design pattern commonly adopted by human experts, and an expressive search space that supports complex topologies. Our algorithm efficiently discovers architectures that outperform a large number of manually designed models for image classification, obtaining top-1 error of 3.6% on CIFAR-10 and 20.3% when transferred to ImageNet, which is competitive with the best existing neural architecture search approaches. We also present results using random search, achieving 0.3% less top-1 accuracy on CIFAR-10 and 0.1% less on ImageNet whilst reducing the search time from 36 hours down to 1 hour. View details
    Relational inductive biases, deep learning, and graph networks
    Peter Battaglia
    Jessica Blake Chandler Hamrick
    Victor Bapst
    Alvaro Sanchez
    Vinicius Zambaldi
    Andrea Tacchetti
    David Raposo
    Adam Santoro
    Ryan Faulkner
    Caglar Gulcehre
    Francis Song
    Andy Ballard
    Justin Gilmer
    Ashish Vaswani
    Kelsey Allen
    Charles Nash
    Victoria Jayne Langston
    Chris Dyer
    Nicolas Heess
    Daan Wierstra
    Matt Botvinick
    Yujia Li
    Razvan Pascanu
    arXiv (2018)
    Preview abstract The purpose of this paper is to explore relational inductive biases in modern AI, especially deep learning, describing a rough taxonomy of existing approaches, and introducing a common mathematical framework for expressing and unifying various approaches. The key theme running through this work is structure—how the world is structured, and how the structure of different computational strategies determines their strengths and weaknesses. View details
    Neural Message Passing for Quantum Chemistry
    Justin Gilmer
    Samuel S. Schoenholz
    Patrick F. Riley
    ICML (2017)
    Preview abstract Supervised learning on molecules has incredible potential to be useful in chemistry, drug discovery, and materials science. Luckily, several promising and closely related neural network models invariant to molecular symmetries have already been described in the literature. These models learn a message passing algorithm and aggregation function to compute a function of their entire input graph. At this point, the next step is to find a particularly effective variant of this general approach and apply it to chemical prediction benchmarks until we either solve them or reach the limits of the approach. In this paper, we reformulate existing models into a single common framework we call Message Passing Neural Networks (MPNNs) and explore additional novel variations within this framework. Using MPNNs we demonstrate state of the art results on an important molecular property prediction benchmark, results we believe are strong enough to justify retiring this benchmark. View details
    Understanding deep learning requires rethinking generalization
    Chiyuan Zhang
    Samy Bengio
    Moritz Hardt
    Benjamin Recht
    ICLR (2017)
    Preview abstract Despite their massive size, successful deep artificial neural networks can exhibit a remarkably small difference between training and test performance. Conventional wisdom attributes small generalization error either to properties of the model family, or to the regularization techniques used during training. Through extensive systematic experiments, we show how these traditional approaches fail to explain why large neural networks generalize well in practice. Specifically, our experiments establish that state-of-the-art convolutional networks for image classification trained with stochastic gradient methods easily fit a random labeling of the training data. This phenomenon is qualitatively unaffected by explicit regularization, and occurs even if we replace the true images by completely unstructured random noise. We corroborate these experimental findings with a theoretical construction showing that simple depth two neural networks already have perfect finite sample expressivity as soon as the number of parameters exceeds the number of data points as it usually does in practice. We interpret our experimental findings by comparison with traditional models. View details
    Parallel WaveNet: Fast High-Fidelity Speech Synthesis
    Aäron van den Oord
    Yazhe Li
    Igor Babuschkin
    Karen Simonyan
    Koray Kavukcuoglu
    George van den Driessche
    Luis Carlos Cobo Rus
    Florian Stimberg
    Norman Casagrande
    Dominik Grewe
    Seb Noury
    Sander Dieleman
    Erich Elsen
    Alexander Graves
    Helen King
    Thomas Walters
    Demis Hassabis
    NA, Google Deepmind, NA (2017)
    Preview abstract The recently-developed WaveNet architecture [27] is the current state of the art in realistic speech synthesis, consistently rated as more natural sounding for many different languages than any previous system. However, because WaveNet relies on sequential generation of one audio sample at a time, it is poorly suited to today’s massively parallel computers, and therefore hard to deploy in a real-time production setting. This paper introduces Probability Density Distillation, a new method for training a parallel feed-forward network from a trained WaveNet with no significant difference in quality. The resulting system is capable of generating high-fidelity speech samples at more than 20 times faster than real-time, and is deployed online by Google Assistant, including serving multiple English and Japanese voices. View details
    Prediction errors of molecular machine learning models lower than hybrid DFT error
    Felix Faber
    Luke Hutchinson
    Huang Bing
    Justin Gilmer
    Sam Schoenholz
    Steven Kearnes
    Patrick Riley
    Anatole von Lilienfeld
    Journal of Chemical Theory and Computation (2017)
    Preview abstract We investigate the impact of choosing regressors and molecular representations for the construction of fast machine learning (ML) models of thirteen electronic ground-state properties of organic molecules. The performance of each regressor/representation/property combination is assessed with learning curves which report approximation errors as a function of training set size. Molecular structures and properties at hybrid density functional theory (DFT) level of theory used for training and testing come from the QM9 database [Ramakrishnan et al, Scientific Data 1 140022 (2014)] and include dipole moment, polarizability, HOMO/LUMO energies and gap, electronic spatial extent, zero point vibrational energy, enthalpies and free energies of atomization, heat capacity and the highest fundamental vibrational frequency. Various representations from the literature have been studied (Coulomb matrix, bag of bonds, BAML and ECFP4, molecular graphs (MG)), as well as newly developed distribution based variants including histograms of distances (HD), and angles (HDA/MARAD), and dihedrals (HDAD). Regressors include linear models (Bayesian ridge regression (BR) and linear regression with elastic net regularization (EN)), random forest (RF), kernel ridge regression (KRR) and two types of neural networks, graph convolutions (GC) and gated graph networks (GG). We present numerical evidence that ML model predictions for all properties can reach an approximation error to DFT which is on par with chemical accuracy. These findings indicate that ML models could be more accurate than DFT if explicitly electron correlated quantum (or experimental) data was provided. View details
    A Neural Transducer
    Navdeep Jaitly
    David Sussillo
    Ilya Sutskever
    Samy Bengio
    NIPS 2016 (2016)
    Preview abstract Sequence-to-sequence models have achieved impressive results on various tasks. However, they are unsuitable for tasks that require incremental predictions to be made as more data arrives or tasks that have long input sequences and output sequences. This is because they generate an output sequence conditioned on an entire input sequence. In this paper, we present a Neural Transducer that can make incremental predictions as more input arrives, without redoing the entire computation. Unlike sequence-to-sequence models, the Neural Transducer computes the next-step distribution conditioned on the partially observed input sequence and the partially generated sequence. At each time step, the transducer can decide to emit zero to many output symbols. The data can be processed using an encoder and presented as input to the transducer. The discrete decision to emit a symbol at every time step makes it difficult to learn with conventional backpropagation. It is however possible to train the transducer by using a dynamic programming algorithm to generate target discrete decisions. Our experiments show that the Neural Transducer works well in settings where it is required to produce output predictions as data come in. We also find that the Neural Transducer performs well for long sequences even when attention mechanisms are not used. View details
    Exploring the limits of language modeling
    Rafal Jozefowicz
    Mike Schuster
    Noam Shazeer
    Google Inc. (2016)
    Preview abstract This paper shows recent advances for large scale neural language modeling, a task central to language understanding. Our goal is to show how well large neural language models can perform on a large LM benchmark corpus, for which we chose the One Billion Word Benchmark. Using various techniques, our best single model significantly improves state-of-the-art perplexity from 51.3 to 30.0, while an ensemble of models sets a new record by improving perplexity from 41.0 to 23.7. View details
    Order matters: Sequence to sequence for sets
    Samy Bengio
    Manjunath Kudlur
    International Conference on Learning Representations (ICLR) (2016)
    Preview abstract Sequences have become first class citizens in supervised learning thanks to the resurgence of recurrent neural networks. Many complex tasks that require mapping from or to a sequence of observations can now be formulated with the sequence-to-sequence (seq2seq) framework which employs the chain rule to efficiently represent the joint probability of sequences. In many cases, however, variable sized inputs and/or outputs might not be naturally expressed as sequences. For instance, it is not clear how to input a set of numbers into a model where the task is to sort them; similarly, we do not know how to organize outputs when they correspond to random variables and the task is to model their unknown joint probability. In this paper, we first show using various examples that the order in which we organize input and/or output data matters significantly when learning an underlying model. We then discuss an extension of the seq2seq framework that goes beyond sequences and handles input sets in a principled way. In addition, we propose a loss which, by searching over possible orders during training, deals with the lack of structure of output sets. We show empirical evidence of our claims regarding ordering, and on the modifications to the seq2seq framework on benchmark language modeling and parsing tasks, as well as two artificial tasks - sorting numbers and estimating the joint probability of unknown graphical models. View details
    Preview abstract Sequence-to-sequence models have achieved impressive results on various tasks. However, they are unsuitable for tasks that require incremental predictions to be made as more data arrives. This is because they generate an output sequence conditioned on an entire input sequence. In this paper, we present a new model that can make incremental predictions as more input arrives, without redoing the entire computation. Unlike sequence-to-sequence models, our method computes the next-step distribution conditioned on the partial input sequence observed and the partial sequence generated. It accomplishes this goal using an encoder recurrent neural network (RNN) that computes features at the same frame rate as the input, and a transducer RNN that operates over blocks of input steps. The transducer RNN extends the sequence produced so far using a local sequence-to-sequence model. During training, our method uses alignment information to generate supervised targets for each block. Approximate alignment is easily available for tasks such as speech recognition, action recognition in videos, etc. During inference (decoding), beam search is used to find the most likely output sequence for an input sequence. This decoding is performed online - at the end of each block, the best candidates from the previous block are extended through the local sequence-to-sequence model. On TIMIT, our online method achieves 19.8% phone error rate (PER). For comparison with published sequence-to-sequence methods, we used a bidirectional encoder and achieved 18.7% PER compared to 17.6% from the best reported sequence-to-sequence model. Importantly, unlike sequence-to-sequence our model is minimally impacted by the length of the input. On artificially created longer utterances, it achieves 20.9% with a unidirectional model, compared to 20% from the best bidirectional sequence-to-sequence models. View details
    Preview abstract The standard recurrent neural network language model (RNNLM) generates sentences one word at a time and does not work from an explicit global sentence representation. In this work, we introduce and study an RNN-based variational autoencoder generative model that incorporates distributed latent representations of entire sentences. This factorization allows it to explicitly model holistic properties of sentences such as style, topic, and high-level syntactic features. Samples from the prior over these sentence representations remarkably produce diverse and well-formed sentences through simple deterministic decoding. By examining paths through this latent space, we are able to generate coherent novel sentences that interpolate between known sentences. We present techniques for solving the difficult learning problem presented by this model, demonstrate its effectiveness in imputing missing words, explore many interesting properties of the model's latent sentence space, and present negative results on the use of the model in language modeling. View details
    Conditional Image Generation with PixelCNN Decoders
    Aäron van den Oord
    Koray Kavukcuoglu
    Alexander Graves
    Advances in Neural Information Processing Systems 29, Curran Associates, Inc. (2016), pp. 4790-4798 (to appear)
    Preview abstract This work explores conditional image generation with a new image density model based on the PixelCNN architecture. The model can be conditioned on any vector, including descriptive labels or tags, or latent embeddings created by other networks. When conditioned on class labels from the ImageNet database, the model is able to generate diverse, realistic scenes representing distinct animals, objects, landscapes and structures. When conditioned on an embedding produced by a convolutional network given a single image of an unseen face, it generates a variety of new portraits of the same person with different facial expressions, poses and lighting conditions. We also show that conditional PixelCNN can serve as a powerful decoder in an image autoencoder. Additionally, the gated convolutional layers in the proposed model improve the log-likelihood of PixelCNN to match the state-ofthe-art performance of PixelRNN on ImageNet, with greatly reduced computational cost. View details
    Preview abstract We present Listen, Attend and Spell (LAS), a neural network that learns to transcribe speech utterances to characters. Unlike traditional DNN-HMM models, this model learns all the components of a speech recognizer jointly. Our system has two components: a listener and a speller. The listener is a pyramidal recurrent network encoder that accepts filter bank spectra as inputs. The speller is an attention based recurrent network decoder that emits characters as outputs. The network produces character sequences without making any independence assumptions between the characters. This is the key improvement of LAS over previous end-to-end CTC models. On a subset of the Google voice search task, LAS achieves a word error rate (WER) of 14.1% without a dictionary or a language model, and 10.3% with language model rescoring over the top 32 beams. By comparison, the state-of-the-art CLDNN-HMM model achieves a WER of 8.0%. View details
    Preview abstract The recent application of RNN encoder-decoder models has resulted in substantial progress in fully data-driven dialogue systems, but evaluation remains a challenge. An adversarial loss could be a way to directly evaluate the extent to which generated dialogue responses sound like they came from a human. This could reduce the need for human evaluation, while more directly evaluating on a generative task. In this work, we investigate this idea by training an RNN to discriminate a dialogue model's samples from human-generated samples. Although we find some evidence this setup could be viable, we also note that many issues remain in its practical application. We discuss both aspects and conclude that future work is warranted. View details
    Preview abstract We describe an LSTM-based model which we call Byte-to-Span (BTS) that reads text as bytes and outputs span annotations of the form [start, length, label] where start positions, lengths, and labels are separate entries in our vocabulary. Because we operate directly on unicode bytes rather than language-specific words or characters, we can analyze text in many languages with a single model. Due to the small vocabulary size, these multilingual models are very compact, but produce results similar to or better than the state-of-the-art in Part-of-Speech tagging and Named Entity Recognition that use only the provided training datasets (no external data sources). Our models are learning “from scratch” in that they do not rely on any elements of the standard pipeline in Natural Language Processing (including tokenization), and thus can run in standalone fashion on raw text. View details
    Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation
    Mike Schuster
    Mohammad Norouzi
    Maxim Krikun
    Qin Gao
    Apurva Shah
    Xiaobing Liu
    Łukasz Kaiser
    Stephan Gouws
    Taku Kudo
    Keith Stevens
    George Kurian
    Nishant Patil
    Wei Wang
    Jason Smith
    Alex Rudnick
    Macduff Hughes
    CoRR, vol. abs/1609.08144 (2016)
    Preview abstract Neural Machine Translation (NMT) is an end-to-end learning approach for automated translation, with the potential to overcome many of the weaknesses of conventional phrase-based translation systems. Unfortunately, NMT systems are known to be computationally expensive both in training and in translation inference. Also, most NMT systems have difficulty with rare words. These issues have hindered NMT's use in practical deployments and services, where both accuracy and speed are essential. In this work, we present GNMT, Google's Neural Machine Translation system, which attempts to address many of these issues. Our model consists of a deep LSTM network with 8 encoder and 8 decoder layers using attention and residual connections. To improve parallelism and therefore decrease training time, our attention mechanism connects the bottom layer of the decoder to the top layer of the encoder. To accelerate the final translation speed, we employ low-precision arithmetic during inference computations. To improve handling of rare words, we divide words into a limited set of common sub-word units ("wordpieces") for both input and output. This method provides a good balance between the flexibility of "character"-delimited models and the efficiency of "word"-delimited models, naturally handles translation of rare words, and ultimately improves the overall accuracy of the system. Our beam search technique employs a length-normalization procedure and uses a coverage penalty, which encourages generation of an output sentence that is most likely to cover all the words in the source sentence. On the WMT'14 English-to-French and English-to-German benchmarks, GNMT achieves competitive results to state-of-the-art. Using a human side-by-side evaluation on a set of isolated simple sentences, it reduces translation errors by an average of 60% compared to Google's phrase-based production system. View details
    Contextual LSTM: A Step towards Hierarchical Language Modeling
    Shalini Ghosh
    Brian Strope
    Scott Roy
    Tom Dean
    Larry Heck
    Workshop on Large­-scale Deep Learning for Data Mining - KDD (2016) (to appear)
    Preview abstract Documents exhibit sequential structure at multiple levels of abstraction (e.g., sentences, paragraphs, sections). These abstractions constitute a natural hierarchy for representing the context in which to infer the meaning of words and larger fragments of text. In this paper, we present CLSTM (Contextual LSTM), an extension of the recurrent neural network LSTM (Long-Short Term Memory) model, where we incorporate contextual features (e.g., topics) into the model. We evaluate CLSTM on three specific NLP tasks: word prediction, next sentence selection, and sentence topic prediction. Results from experiments run on two corpora, English documents in Wikipedia and a subset of articles from a recent snapshot of English Google News, indicate that using both words and topics as features improves performance of the CLSTM models over baseline LSTM models for these tasks. For example on the next sentence selection task, we get relative accuracy improvements of 21% for the Wikipedia dataset and 18% for the Google News dataset. This clearly demonstrates the significant benefit of using context appropriately in natural language (NL) tasks. This has implications for a wide variety of NL applications like question answering, sentence completion, paraphrase generation, and next utterance prediction in dialog systems. View details
    WaveNet: A Generative Model for Raw Audio
    Aäron van den Oord
    Sander Dieleman
    Karen Simonyan
    Alexander Graves
    Koray Kavukcuoglu
    Arxiv (2016)
    Preview abstract This paper introduces WaveNet, a deep generative neural network trained end-to-end to model raw audio waveforms, which can be applied to text-to-speech and music generation. Current approaches to text-to-speech are focused on non-parametric, example-based generation (which stitches together short audio signal segments from a large training set), and parametric, model-based generation (in which a model generates acoustic features synthesized into a waveform with a vocoder). In contrast, we show that directly generating wideband audio signals at tens of thousands of samples per second is not only feasible, but also achieves results that significantly outperform the prior art. A single trained WaveNet can be used to generate different voices by conditioning on the speaker identity. We also show that the same approach can be used for music audio generation and speech recognition. View details
    Multi-task Sequence to Sequence Learning
    Ilya Sutskever
    Lukasz Kaiser
    International Conference on Learning Representations (2016)
    Preview abstract Sequence to sequence learning has recently emerged as a new paradigm in supervised learning. To date, most of its applications focused on only one task and not much work explored this framework for multiple tasks. This paper examines three multi-task learning (MTL) settings for sequence to sequence models: (a) the oneto-many setting - where the encoder is shared between several tasks such as machine translation and syntactic parsing, (b) the many-to-one setting - useful when only the decoder can be shared, as in the case of translation and image caption generation, and (c) the many-to-many setting - where multiple encoders and decoders are shared, which is the case with unsupervised objectives and translation. Our results show that training on a small amount of parsing and image caption data can improve the translation quality between English and German by up to 1.5 BLEU points over strong single-task baselines on the WMT benchmarks. Furthermore, we have established a new state-of-the-art result in constituent parsing with 93.0 F1. Lastly, we reveal interesting properties of the two unsupervised learning objectives, autoencoder and skip-thought, in the MTL context: autoencoder helps less in terms of perplexities but more on BLEU scores compared to skip-thought. View details
    Listen, Attend and Spell
    William Chan
    Navdeep Jaitly
    CoRR, vol. abs/1508.01211 (2015)
    Preview abstract We present Listen, Attend and Spell (LAS), a neural network that learns to transcribe speech utterances to characters. Unlike traditional DNN-HMM models, this model learns all the components of a speech recognizer jointly. Our system has two components: a listener and a speller. The listener is a pyramidal recurrent network encoder that accepts filter bank spectra as inputs. The speller is an attention-based recurrent network decoder that emits characters as outputs. The network produces character sequences without making any independence assumptions between the characters. This is the key improvement of LAS over previous end-to-end CTC models. On a subset of the Google voice search task, LAS achieves a word error rate (WER) of 14.1% without a dictionary or a language model, and 10.3% with language model rescoring over the top 32 beams. By comparison, the state-of-the-art CLDNN-HMM model achieves a WER of 8.0%. View details
    Sentence Compression by Deletion with LSTMs
    Lukasz Kaiser
    Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing (EMNLP'15)
    Preview abstract We present an LSTM approach to deletion-based sentence compression where the task is to translate a sentence into a sequence of zeros and ones, corresponding to token deletion decisions. We demonstrate that even the most basic version of the system, which is given no syntactic information (no PoS or NE tags, or dependencies) or desired compression length, performs surprisingly well: around 30% of the compressions from a large test set could be regenerated. We compare the LSTM system with a competitive baseline which is trained on the same amount of data but is additionally provided with all kinds of linguistic features. In an experiment with human raters the LSTM-based model outperforms the baseline achieving 4.5 in readability and 3.8 in informativeness. View details
    Qualitatively Characterizing Neural Network Optimization Problems
    Ian Goodfellow
    Andrew Saxe
    International Conference on Learning Representations (2015)
    Preview abstract Training neural networks involves solving large-scale non-convex optimization problems. This task has long been believed to be extremely difficult, with fear of local minima and other obstacles motivating a variety of schemes to improve optimization, such as unsupervised pretraining. However, modern neural networks are able to achieve negligible training error on complex tasks, using only direct training with stochastic gradient descent. We introduce a simple analysis technique to look for evidence that such networks are overcoming local optima. We find that, in fact, on a straight path from initialization to solution, a variety of state of the art neural networks never encounter any significant obstacles. View details
    Pointer Networks
    Meire Fortunato
    Navdeep Jaitly
    NIPS (2015), pp. 2692-2700
    Preview abstract We introduce a new neural architecture to learn the conditional probability of an output sequence with elements that are discrete tokens corresponding to positions in an input sequence. Such problems cannot be trivially addressed by existent approaches such as sequence-to-sequence [1] and Neural Turing Machines [2], because the number of target classes in each step of the output depends on the length of the input, which is variable. Problems such as sorting variable sized sequences, and various combinatorial optimization problems belong to this class. Our model solves the problem of variable size output dictionaries using a recently proposed mechanism of neural attention. It differs from the previous attention attempts in that, instead of using attention to blend hidden units of an encoder to a context vector at each decoder step, it uses attention as a pointer to select a member of the input sequence as the output. We call this architecture a Pointer Net (Ptr-Net). We show Ptr-Nets can be used to learn approximate solutions to three challenging geometric problems – finding planar convex hulls, computing Delaunay triangulations, and the planar Travelling Salesman Problem – using training examples alone. Ptr-Nets not only improve over sequence-to-sequence with input attention, but also allow us to generalize to variable size output dictionaries. We show that the learnt models generalize beyond the maximum lengths they were trained on. We hope our results on these tasks will encourage a broader exploration of neural learning for discrete problems. View details
    Preview abstract Conversational modeling is an important task in natural language understanding and machine intelligence. Although previous approaches exist, they are often restricted to specific domains (e.g., booking an airline ticket) and require hand-crafted rules. In this paper, we present a simple approach for this task which uses the recently proposed sequence to sequence framework. Our model converses by predicting the next sentence given the previous sentence or sentences in a conversation. The strength of our model is that it can be trained end-to-end and thus requires much fewer hand-crafted rules. We find that this straightforward model can generate simple conversations given a large conversational training dataset. Our preliminary results suggest that, despite optimizing the wrong objective function, the model is able to converse well. It is able extract knowledge from both a domain specific dataset, and from a large, noisy, and general domain dataset of movie subtitles. On a domain-specific IT helpdesk dataset, the model can find a solution to a technical problem via conversations. On a noisy open-domain movie transcript dataset, the model can perform simple forms of common sense reasoning. As expected, we also find that the lack of consistency is a common failure mode of our model. View details
    Preview abstract Neural Machine Translation (NMT) is a new approach to machine translation that has shown promising results that are comparable to traditional approaches. A significant weakness in conventional NMT systems is their inability to correctly translate very rare words: end-to-end NMTs tend to have relatively small vocabularies with a single unk symbol that represents every possible out-of-vocabulary (OOV) word. In this paper, we propose and implement an effective technique to address this problem. We train an NMT system on data that is augmented by the output of a word alignment algorithm, allowing the NMT system to emit, for each OOV word in the target sentence, the position of its corresponding word in the source sentence. This information is later utilized in a post-processing step that translates every OOV word using a dictionary. Our experiments on the WMT’14 English to French translation task show that this method provides a substantial improvement of up to 2.8 BLEU points over an equivalent NMT system that does not use this technique. With 37.5 BLEU points, our NMT system is the first to surpass the best result achieved on a WMT’14 contest task. View details
    Beyond Short Snippets: Deep Networks for Video Classification
    Joe Yue-Hei Ng
    Matthew Hausknecht
    Rajat Monga
    Computer Vision and Pattern Recognition (2015)
    Preview abstract Convolutional neural networks (CNNs) have been extensively applied for image recognition problems giving state-of-the-art results on recognition, detection, segmentation and retrieval. In this work we propose and evaluate several deep neural network architectures to combine image information across a video over longer time periods than previously attempted. We propose two methods capable of handling full length videos. The first method explores various convolutional temporal feature pooling architectures, examining the various design choices which need to be made when adapting a CNN for this task. The second proposed method explicitly models the video as an ordered sequence of frames. For this purpose we employ a recurrent neural network that uses Long Short-Term Memory (LSTM) cells which are connected to the output of the underlying CNN. Our best networks exhibit significant performance improvements over previously published results on the Sports 1 million dataset (73.1% vs. 60.9%) and the UCF-101 datasets with (88.6% vs. 88.0%) and without additional optical flow information (82.6% vs. 72.8%). View details
    Scheduled Sampling for Sequence Prediction with Recurrent Neural Networks
    Samy Bengio
    Navdeep Jaitly
    Noam M. Shazeer
    Advances in Neural Information Processing Systems, NIPS (2015)
    Preview abstract Recurrent Neural Networks can be trained to produce sequences of tokens given some input, as exemplified by recent results in machine translation and image captioning. The current approach to training them consists of maximizing the likelihood of each token in the sequence given the current (recurrent) state and the previous token. At inference, the unknown previous token is then replaced by a token generated by the model itself. This discrepancy between training and inference can yield errors that can accumulate quickly along the generated sequence. We propose a curriculum learning strategy to gently change the training process from a fully guided scheme using the true previous token, towards a less guided scheme which mostly uses the generated token instead. Experiments on several sequence prediction tasks show that this approach yields significant improvements. Moreover, it was used successfully in our winning entry to the MSCOCO image captioning challenge, 2015. View details
    Show and tell: A neural image caption generator
    Alexander Toshev
    Samy Bengio
    Computer Vision and Pattern Recognition (2015)
    Preview abstract Automatically describing the content of an image is a fundamental problem in artificial intelligence that connects computer vision and natural language processing. In this paper, we present a generative model based on a deep recurrent architecture that combines recent advances in computer vision and machine translation and that can be used to generate natural sentences describing an image. The model is trained to maximize the likelihood of the target description sentence given the training image. Experiments on several datasets show the accuracy of the model and the fluency of the language it learns solely from image descriptions. Our model is often quite accurate, which we verify both qualitatively and quantitatively. For instance, while the current state-of-the-art BLEU score (the higher the better) on the Pascal dataset is 25, our approach yields 59, to be compared to human performance around 69. We also show BLEU-1 score improvements on Flickr30k, from 56 to 66, and on SBU, from 19 to 28. Lastly, on the newly released COCO dataset, we achieve a BLEU-4 of 27.7, which is the current state-of-the-art. View details
    Towards Principled Unsupervised Learning
    Ilya Sutskever
    Rafal Jozefowicz
    Karol Gregor
    Danilo Rezende
    Tim Lillicrap
    Google Inc. (2015)
    Preview abstract General unsupervised learning is a long-standing conceptual problem in machine learning. Supervised learning is successful because it can be solved by the minimization of the training error cost function. Unsupervised learning is not as successful, because the unsupervised objective may be unrelated to the supervised task of interest. For an example, density modelling and reconstruction have often been used for unsupervised learning, but they did not produced the sought-after performance gains, because they have no knowledge of the supervised tasks. In this paper, we present an unsupervised cost function which we name the Output Distribution Matching (ODM) cost, which measures a divergence between the distribution of predictions and distributions of labels. The ODM cost is appealing because it is consistent with the supervised cost in the following sense: a perfect supervised classifier is also perfect according to the ODM cost. Therefore, by aggressively optimizing the ODM cost, we are almost guaranteed to improve our supervised performance whenever the space of possible predictions is exponentially large. We demonstrate that the ODM cost works well on number of small and semi-artificial datasets using no (or almost no) labelled training cases. Finally, we show that the ODM cost can be used for one-shot domain adaptation, which allows the model to classify inputs that differ from the input distribution in significant ways without the need for prior exposure to the new domain. View details
    Grammar as a Foreign Language
    Lukasz Kaiser
    Terry Koo
    Ilya Sutskever
    Geoffrey Hinton
    NIPS (2015)
    Preview abstract Syntactic constituency parsing is a fundamental problem in natural language processing and has been the subject of intensive research and engineering for decades. As a result, the most accurate parsers are domain specific, complex, and inefficient. In this paper we show that the domain agnostic attention-enhanced sequence-to-sequence model achieves state-of-the-art results on the most widely used syntactic constituency parsing dataset, when trained on a large synthetic corpus that was annotated using existing parsers. It also matches the performance of standard parsers when trained only on a small human-annotated dataset, which shows that this model is highly data-efficient, in contrast to sequence-to-sequence models without the attention mechanism. Our parser is also fast, processing over a hundred sentences per second with an unoptimized CPU implementation. View details
    TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems
    Ashish Agarwal
    Eugene Brevdo
    Craig Citro
    Matthieu Devin
    Ian Goodfellow
    Andrew Harp
    Geoffrey Irving
    Yangqing Jia
    Rafal Jozefowicz
    Lukasz Kaiser
    Manjunath Kudlur
    Dan Mané
    Rajat Monga
    Chris Olah
    Mike Schuster
    Jonathon Shlens
    Benoit Steiner
    Ilya Sutskever
    Kunal Talwar
    Paul Tucker
    Vijay Vasudevan
    Pete Warden
    Yuan Yu
    Xiaoqiang Zheng
    tensorflow.org (2015)
    Preview abstract TensorFlow is an interface for expressing machine learning algorithms, and an implementation for executing such algorithms. A computation expressed using TensorFlow can be executed with little or no change on a wide variety of heterogeneous systems, ranging from mobile devices such as phones and tablets up to large-scale distributed systems of hundreds of machines and thousands of computational devices such as GPU cards. The system is flexible and can be used to express a wide variety of algorithms, including training and inference algorithms for deep neural network models, and it has been used for conducting research and for deploying machine learning systems into production across more than a dozen areas of computer science and other fields, including speech recognition, computer vision, robotics, information retrieval, natural language processing, geographic information extraction, and computational drug discovery. This paper describes the TensorFlow interface and an implementation of that interface that we have built at Google. The TensorFlow API and a reference implementation were released as an open-source package under the Apache 2.0 license in November, 2015 and are available at www.tensorflow.org. View details
    Preview abstract Both Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) have shown improvements over Deep Neural Networks (DNNs) across a wide variety of speech recognition tasks. CNNs, LSTMs and DNNs are complementary in their modeling capabilities, as CNNs are good at reducing frequency variations, LSTMs are good at temporal modeling, and DNNs are appropriate for mapping features to a more separable space. In this paper, we take advantage of the complementarity of CNNs, LSTMs and DNNs by combining them into one unified architecture. We explore the proposed architecture, which we call CLDNN, on a variety of large vocabulary tasks, varying from 200 to 2,000 hours. We find that the CLDNN provides a 4-6% relative improvement in WER over an LSTM, the strongest of the three individual models. View details
    Distilling the Knowledge in a Neural Network
    Geoffrey Hinton
    NIPS Deep Learning and Representation Learning Workshop (2015)
    Preview abstract A very simple way to improve the performance of almost any machine learning algorithm is to train many different models on the same data and then to average their predictions. Unfortunately, making predictions using a whole ensemble of models is cumbersome and may be too computationally expensive to allow deployment to a large number of users, especially if the individual models are large neural nets. Caruana and his collaborators have shown that it is possible to compress the knowledge in an ensemble into a single model which is much easier to deploy and we develop this approach further using a different compression technique. We achieve some surprising results on MNIST and we show that we can significantly improve the acoustic model of a heavily used commercial system by distilling the knowledge in an ensemble of models into a single model. We also introduce a new type of ensemble composed of one or more full models and many specialist models which learn to distinguish fine-grained classes that the full models confuse. Unlike a mixture of experts, these specialist models can be trained rapidly and in parallel. View details
    Recurrent Neural Network Regularization
    Wojciech Zaremba
    Ilya Sutskever
    Google Inc. (2014)
    Preview abstract We present a simple regularization technique for Recurrent Neural Networks (RNNs) with Long Short-Term Memory (LSTM) units. Dropout, the most successful technique for regularizing neural networks, does not work well with RNNs and LSTMs. In this paper, we show how to correctly apply dropout to LSTMs, and show that it substantially reduces overfitting on a variety of tasks. These tasks include language modeling, speech recognition, image caption generation, and machine translation. View details
    Preview abstract Deep Neural Networks (DNNs) are powerful models that have achieved excellent performance on difficult learning tasks. Although DNNs work well whenever large labeled training sets are available, they cannot be used to map sequences to sequences. In this paper, we present a general end-to-end approach to sequence learning that makes minimal assumptions on the sequence structure. Our method uses a multilayered Long Short-Term Memory (LSTM) to map the input sequence to a vector of a fixed dimensionality, and then another deep LSTM to decode the target sequence from the vector. Our main result is that on an English to French translation task from the WMT-14 dataset, the translations produced by the LSTM achieve a BLEU score of 34.8 on the entire test set, where the LSTM's BLEU score was penalized on out-of-vocabulary words. Additionally, the LSTM did not have difficulty on long sentences. For comparison, a phrase-based SMT system achieves a BLEU score of 33.3 on the same dataset. When we used the LSTM to rerank the 1000 hypotheses produced by the aforementioned SMT system, its BLEU score increases to 36.5, which is close to the previous state of the art. The LSTM also learned sensible phrase and sentence representations that are sensitive to word order and are relatively invariant to the active and the passive voice. Finally, we found that reversing the order of the words in all source sentences (but not target sentences) improved the LSTM's performance markedly, because doing so introduced many short term dependencies between the source and the target sentence which made the optimization problem easier. View details
    Preview abstract We recently showed that Long Short-Term Memory (LSTM) recurrent neural networks (RNNs) outperform state-of-the-art deep neural networks (DNNs) for large scale acoustic modeling where the models were trained with the cross-entropy (CE) criterion. It has also been shown that sequence discriminative training of DNNs initially trained with the CE criterion gives significant improvements. In this paper, we investigate sequence discriminative training of LSTM RNNs in a large scale acoustic modeling task. We train the models in a distributed manner using asynchronous stochastic gradient descent optimization technique. We compare two sequence discriminative criteria -- maximum mutual information and state-level minimum Bayes risk, and we investigate a number of variations of the basic training strategy to better understand issues raised by both the sequential model, and the objective function. We obtain significant gains over the CE trained LSTM RNN model using sequence discriminative training techniques. View details
    Recurrent Neural Networks for Noise Reduction in Robust ASR
    Andrew Maas
    Tyler M. O’Neil
    Patrick Nguyen
    Andrew Y. Ng
    INTERSPEECH (2012)
    Preview abstract Recent work on deep neural networks as acoustic models for automatic speech recognition (ASR) have demonstrated substantial performance improvements. We introduce a model which uses a deep recurrent auto encoder neural network to denoise input features for robust ASR. The model is trained on stereo (noisy and clean) audio features to predict clean features given noisy input. The model makes no assumptions about how noise affects the signal, nor the existence of distinct noise environments. Instead, the model can learn to model any type of distortion or additive noise given sufficient training data. We demonstrate the model is competitive with existing feature denoising approaches on the Aurora2 task, and outperforms a tandem approach where deep networks are used to predict phoneme posteriors directly. View details
    No Results Found