Machine learning classifiers are known to be vulnerable to inputs maliciously constructed by adversaries to force misclassification. Such adversarial examples have been extensively studied in the context of computer vision applications. In this work, we show adversarial attacks are also effective when targeting neural network policies in reinforcement learning. Specifically, we show existing adversarial example crafting techniques can be used to significantly degrade test-time performance of trained policies. Our threat model considers adversaries capable of introducing small perturbations to the raw input of the policy. We characterize the degree of vulnerability across tasks and training algorithms, for a subclass of adversarial-example attacks in white-box and black-box settings. Regardless of the learned task or training algorithm, we observe a significant drop in performance, even with small adversarial perturbations that do not interfere with human perception. Videos are available at http://rll.berkeley.edu/adversarial.

Inference and learning for probabilistic generative networks is often very challenging and typically prevents scalability to as large networks as used for deep discriminative approaches. To obtain efficiently trainable, large-scale and well performing generative networks for semi-supervised learning, we here combine two recent developments: a neural network reformulation of hierarchical Poisson mixtures (Neural Simpletrons), and a novel truncated variational EM approach (TV-EM). TV-EM provides theoretical guarantees for learning in generative networks, and its application to Neural Simpletrons results in particularly compact, yet approximately optimal, modifications of learning equations. If applied to standard benchmarks, we empirically find, that learning converges in fewer EM iterations, that the complexity per EM iteration is reduced, and that final likelihood values are higher on average. For the task of classification on data sets with few labels, learning improvements result in consistently lower error rates if compared to applications without truncation. Experiments on the MNIST data set herein allow for comparison to standard and state-of-the-art models in the semi-supervised setting. Further experiments on the NIST SD19 data set show the scalability of the approach when a manifold of additional unlabeled data is available.

This paper presents a novel model for multimodal learning based on gated neural networks. The Gated Multimodal Unit (GMU) model is intended to be used as an internal unit in a neural network architecture whose purpose is to find an intermediate representation based on a combination of data from different modalities. The GMU learns to decide how modalities influence the activation of the unit using multiplicative gates. It was evaluated on a multilabel scenario for genre classification of movies using the plot and the poster. The GMU improved the macro f-score performance of single-modality approaches and outperformed other fusion strategies, including mixture of experts models. Along with this work, the MM-IMDb dataset is released which, to the best of our knowledge, is the largest publicly available multimodal dataset for genre prediction on movies.

Sequence-to-sequence models rely on a fixed decomposition of the target sequences into a sequence of tokens that may be words, word-pieces or characters. The choice of these tokens and the decomposition of the target sequences into a sequence of tokens is often static, and independent of the input, output data domains. This can potentially lead to a sub-optimal choice of token dictionaries, as the decomposition is not informed by the particular problem being solved. In this paper we present Latent Sequence Decompositions (LSD), a framework in which the decomposition of sequences into constituent tokens is learnt during the training of the model. The decomposition depends both on the input sequence and on the output sequence. In LSD, during training, the model samples decompositions incrementally, from left to right by locally sampling between valid extensions.

Learning to learn by gradient descent by gradient descent

Don't let the fun, casual name mislead you. Chatbots--software that you can "chat with"--have serious implications for the business world. Though many businesses have already considered their use for customer service purposes, a chatbot's internal applications could be invaluable on a larger scale. For instance, chatbots could help employees break down siloes and provide targeted data to fuel every department. This digital transformation is happening, even in organizational structures that face challenges with other formats of real-time communication.

Did your voice give it away? US start-up Canary Speech is developing deep-learning algorithms to detect if people have neurological conditions like Parkinson's or Alzheimer's disease just by listening to the sound of their voice. And it's found a controversial source of audio data to train its algorithms on: phone calls to a health insurer. The health insurer – which Canary Speech would not name but says is "a very large American healthcare and insurance provider" – has provided the company with hundreds of millions of phone calls that have been collected over the past 15 years and are labelled with information about the speaker's medical history and demographic background. Using this data, the company says its algorithms could pick up on vocal cues that distinguish someone with a particular condition from someone without that condition.

Perhaps the most positive technical theme of 2016 was the long-delayed triumph of artificial intelligence, machine learning, and in particular deep learning. In this article we'll discuss what that means and how you might make use of deep learning yourself. Perhaps you noticed in the fall of 2016 that Google Translate suddenly went from producing, on the average, word salad with a vague connection to the original language to emitting polished, coherent sentences more often than not -- at least for supported language pairs, such as English-French, English-Chinese, and English-Japanese. That dramatic improvement was the result of a nine-month concerted effort by the Google Brain and Google Translate teams to revamp Translate from using its old phrase-based statistical machine translation algorithms to working with a neural network trained with deep learning and word embeddings employing Google's TensorFlow framework. The researchers working on the conversion had access to a huge corpus of translations from which to train their networks, but they soon discovered that they needed thousands of GPUs for training and would have to create a new kind of chip, a Tensor Processing Unit (TPU), to run Translate on their trained neural networks at scale.

Intel has further detailed their Lake Crest chip that will be aiming at the deep neural network sector. The new chip will be based around the Nervana platform which would deliver an unprecedented amount of compute density in silicon that delivers more raw power than modern GPUs. With the rise of AI learning in the tech industry, GPU makers such as NVIDIA and AMD have made chips that are specifically designed for DNN (Deep Neural Network) workloads. Intel wants to enter this ground with the Lake Crest silicon which is said to deliver more raw power than the fastest DNN GPUs available today. The chip will feature technology developed by the deep-learning startup, Nervana.

A new report from market research firm Tractica forecasts that the annual global revenue for artificial intelligence products and services will grow from 643.7 million in 2016 to $36.8 billion by 2025, a 57-fold increase over that time period. As such, it represents the fastest growing segment of any size in the IT sector. According to the report, "AI has applications and use cases in almost every industry vertical and is considered the next big technological shift, similar to past shifts like the industrial revolution, the computer age, and the smartphone revolution." Even ignoring that bit of hyperbole, a good case can be made that artificial intelligence certainly has the potential to upend a lot of industries, government activities and consumer behavior. In its report, Tactica identifies 27 verticals that are employing AI technologies today or soon will be.