Author: Mu Li, Principal Scientist at Amazon Translated from: https://zh.mxnet.io/blog/gluon-cv Someone once asked me what was the hardest thing to do when developing MXNet. I would not hesitate to say that replicating experimental results from papers is the most difficult part. The heroes in the three examples above are top-level researchers in the field of deep learning, but it is still very easy to spend a lot of precious time on some subtle experimental details. A model usually has tens to hundreds of layers and can take several hours to train.

Microsoft wants to make finding just the right image online less of a crapshoot by using deep learning to help its Bing Image Search engine return more relevant results to users. Deep learning is a computationally intensive subset of machine learning that enables advanced analytical workloads. Using neural networks modeled after the human brain, deep learning systems can better mimic how people recognize patterns and process information. Peeling back the curtain a bit on the search engine's internal workings, the Bing Image Search Relevance Team at Microsoft revealed that they are using deep learning to unlock information held between the pixels of the billions of pictures found online. Microsoft's artificial intelligence (AI) searches for correlations between images and search queries mapped to semantic spaces used to derive meaning from data, even if the web pages containing those images lack any sort of text descriptors.

Build an AI that combines the power of Data Science, Machine Learning and Deep Learning to create powerful AI for Real-World applications. You will also have the chance to understand the story behind Artificial Intelligence. Completely understand the relationship between reinforcement learning and psychology and on a technical level. Apply gradient-based supervised machine learning methods to reinforcement learning and implement 17 different reinforcement learning algorithms.

We present an application of deep generative models in the context of partial-differential equation (PDE) constrained inverse problems. We combine a generative adversarial network (GAN) representing an a priori model that creates subsurface geological structures and their petrophysical properties, with the numerical solution of the PDE governing the propagation of acoustic waves within the earth's interior. We perform Bayesian inversion using an approximate Metropolis-adjusted Langevin algorithm (MALA) to sample from the posterior given seismic observations. Gradients with respect to the model parameters governing the forward problem are obtained by solving the adjoint of the acoustic wave equation. Gradients of the mismatch with respect to the latent variables are obtained by leveraging the differentiable nature of the deep neural network used to represent the generative model. We show that approximate MALA sampling allows efficient Bayesian inversion of model parameters obtained from a prior represented by a deep generative model, obtaining a diverse set of realizations that reflect the observed seismic response.

Many parametric statistical models are not properly normalised and only specified up to an intractable partition function, which renders parameter estimation difficult. Examples of unnormalised models are Gibbs distributions, Markov random fields, and neural network models in unsupervised deep learning. In previous work, the estimation principle called noise-contrastive estimation (NCE) was introduced where unnormalised models are estimated by learning to distinguish between data and auxiliary noise. An open question is how to best choose the auxiliary noise distribution. We here propose a new method that addresses this issue. The proposed method shares with NCE the idea of formulating density estimation as a supervised learning problem but in contrast to NCE, the proposed method leverages the observed data when generating noise samples. The noise can thus be generated in a semi-automated manner. We first present the underlying theory of the new method, show that score matching emerges as a limiting case, validate the method on continuous and discrete valued synthetic data, and show that we can expect an improved performance compared to NCE when the data lie in a lower-dimensional manifold. Then we demonstrate its applicability in unsupervised deep learning by estimating a four-layer neural image model.

In neural abstractive summarization, the conventional sequence-to-sequence (seq2seq) model often suffers from repetition and semantic irrelevance. To tackle the problem, we propose a global encoding framework, which controls the information flow from the encoder to the decoder based on the global information of the source context. It consists of a convolutional gated unit to perform global encoding to improve the representations of the source-side information. Evaluations on the LCSTS and the English Gigaword both demonstrate that our model outperforms the baseline models, and the analysis shows that our model is capable of reducing repetition.

A great proportion of sequence-to-sequence (Seq2Seq) models for Neural Machine Translation (NMT) adopt Recurrent Neural Network (RNN) to generate translation word by word following a sequential order. As the studies of linguistics have proved that language is not linear word sequence but sequence of complex structure, translation at each step should be conditioned on the whole target-side context. To tackle the problem, we propose a new NMT model that decodes the sequence with the guidance of its structural prediction of the context of the target sequence. Our model generates translation based on the structural prediction of the target-side context so that the translation can be freed from the bind of sequential order. Experimental results demonstrate that our model is more competitive compared with the state-of-the-art methods, and the analysis reflects that our model is also robust to translating sentences of different lengths and it also reduces repetition with the instruction from the target-side context for decoding.

In traditional reinforcement learning, an agent maximizes the reward collected during its interaction with the environment by approximating the optimal policy through the estimation of value functions. Typically, given a state s and action a, the corresponding value is the expected discounted sum of rewards. The optimal action is then chosen to be the action a with the largest value estimated by value function. However, recent developments have shown both theoretical and experimental evidence of superior performance when value function is replaced with value distribution in context of deep Q learning [1]. In this paper, we develop a new algorithm that combines advantage actor-critic with value distribution estimated by quantile regression. We evaluated this new algorithm, termed Distributional Advantage Actor-Critic (DA2C or QR-A2C) on a variety of tasks, and observed it to achieve at least as good as baseline algorithms, and outperforming baseline in some tasks with smaller variance and increased stability.

Recent works have shown that, by allowing some inaccuracy when training deep neural networks, not only the training performance but also the accuracy of the model can be improved. Neural networks are capable of learning output functions that can change wildly with small changes in input. Adding noise to inputs randomly is like telling the network to not change the output in a ball around your exact input. By limiting the amount of information in a network, we force it to learn compact representations of input features. RL is an area of machine learning that assumes there is an agent situated in an environment.

Three-year olds are known for a long list of bad habits: biting other kids, throwing toys at their mom and answering every question with "no." Despite those irrational behaviours, they are also smart. Show a three-year-old girl a van, a truck and a car, and she will quickly learn to identify the three types of vehicles. Digvir Jayas, vice-president of research at the University of Manitoba and grain storage expert, said computers aren't as smart as three- year olds, at least when it comes to computer vision and identifying objects. But scientists are now teaching computers to think like a three-year old, so the machines can see the differences between one object and another.