Multi-channel speech enhancement with ad-hoc sensors has been a challenging task. Speech model guided beamforming algorithms are able to recover natural sounding speech, but the speech models tend to be oversimplified or the inference would otherwise be too complicated. On the other hand, deep learning based enhancement approaches are able to learn complicated speech distributions and perform efficient inference, but they are unable to deal with variable number of input channels. Also, deep learning approaches introduce a lot of errors, particularly in the presence of unseen noise types and settings. We have therefore proposed an enhancement framework called DEEPBEAM, which combines the two complementary classes of algorithms. DEEPBEAM introduces a beamforming filter to produce natural sounding speech, but the filter coefficients are determined with the help of a monaural speech enhancement neural network. Experiments on synthetic and real-world data show that DEEPBEAM is able to produce clean, dry and natural sounding speech, and is robust against unseen noise.

Despite the recent successes of deep neural networks in various fields such as image and speech recognition, natural language processing, and reinforcement learning, we still face big challenges in bringing the power of numeric optimization to symbolic reasoning. Researchers have proposed different avenues such as neural machine translation for proof synthesis, vectorization of symbols and expressions for representing symbolic patterns, and coupling of neural back-ends for dimensionality reduction with symbolic front-ends for decision making. However, these initial explorations are still only point solutions, and bear other shortcomings such as lack of correctness guarantees. In this paper, we present our approach of casting symbolic reasoning as games, and directly harnessing the power of deep reinforcement learning in the style of Alpha(Go) Zero on symbolic problems. Using the Boolean Satisfiability (SAT) problem as showcase, we demonstrate the feasibility of our method, and the advantages of modularity, efficiency, and correctness guarantees.

ConvNets, through their architecture, only enforce invariance to translation. In this paper, we introduce a new class of deep convolutional architectures called Non-Parametric Transformation Networks (NPTNs) which can learn \textit{general} invariances and symmetries directly from data. NPTNs are a natural generalization of ConvNets and can be optimized directly using gradient descent. Unlike almost all previous works in deep architectures, they make no assumption regarding the structure of the invariances present in the data and in that aspect are flexible and powerful. We also model ConvNets and NPTNs under a unified framework called Transformation Networks (TN), which yields a better understanding of the connection between the two. We demonstrate the efficacy of NPTNs on data such as MNIST and CIFAR10 where they outperform ConvNet baselines with the same number of parameters. We show it is more effective than ConvNets in modelling symmetries from data, without the explicit knowledge of the added arbitrary nuisance transformations. Finally, we replace ConvNets with NPTNs within Capsule Networks and show that this enables Capsule Nets to perform even better.

Hyperparameter tuning is one of the most time-consuming workloads in deep learning. State-of-the-art optimizers, such as AdaGrad, RMSProp and Adam, reduce this labor by adaptively tuning an individual learning rate for each variable. Recently researchers have shown renewed interest in simpler methods like momentum SGD as they may yield better test metrics. Motivated by this trend, we ask: can simple adaptive methods based on SGD perform as well or better? We revisit the momentum SGD algorithm and show that hand-tuning a single learning rate and momentum makes it competitive with Adam. We then analyze its robustness to learning rate misspecification and objective curvature variation. Based on these insights, we design YellowFin, an automatic tuner for momentum and learning rate in SGD. YellowFin optionally uses a negative-feedback loop to compensate for the momentum dynamics in asynchronous settings on the fly. We empirically show that YellowFin can converge in fewer iterations than Adam on ResNets and LSTMs for image recognition, language modeling and constituency parsing, with a speedup of up to 3.28x in synchronous and up to 2.69x in asynchronous settings.

Catastrophic forgetting occurs when a neural network loses the information learned in a previous task after training on subsequent tasks. This problem remains a hurdle for artificial intelligence systems with sequential learning capabilities. In this paper, we propose a task-based hard attention mechanism that preserves previous tasks' information without affecting the current task's learning. A hard attention mask is learned concurrently to every task, through stochastic gradient descent, and previous masks are exploited to condition such learning. We show that the proposed mechanism is effective for reducing catastrophic forgetting, cutting current rates by 45 to 80%. We also show that it is robust to different hyperparameter choices, and that it offers a number of monitoring capabilities. The approach features the possibility to control both the stability and compactness of the learned knowledge, which we believe makes it also attractive for online learning or network compression applications.

Teachers intentionally pick the most informative examples to show their students. However, if the teacher and student are neural networks, the examples that the teacher network learns to give, although effective at teaching the student, are typically uninterpretable. We show that training the student and teacher iteratively, rather than jointly, can produce interpretable teaching strategies. We evaluate interpretability by (1) measuring the similarity of the teacher's emergent strategies to intuitive strategies in each domain and (2) conducting human experiments to evaluate how effective the teacher's strategies are at teaching humans. We show that the teacher network learns to select or generate interpretable, pedagogical examples to teach rule-based, probabilistic, boolean, and hierarchical concepts.

Planning problems in partially observable environments cannot be solved directly with convolutional networks and require some form of memory. But, even memory networks with sophisticated addressing schemes are unable to learn intelligent reasoning satisfactorily due to the complexity of simultaneously learning to access memory and plan. To mitigate these challenges we introduce the Memory Augmented Control Network (MACN). The proposed network architecture consists of three main parts. The first part uses convolutions to extract features and the second part uses a neural network-based planning module to pre-plan in the environment. The third part uses a network controller that learns to store those specific instances of past information that are necessary for planning. The performance of the network is evaluated in discrete grid world environments for path planning in the presence of simple and complex obstacles. We show that our network learns to plan and can generalize to new environments.

But six months after China seemed to mimic that Obama-era road map, A.I. experts in industry and academia in the United States say that the Trump White House has done little to follow through on the previous administration's economic call to arms. "We are still waiting on the White House to provide some direction" on how to respond to the competition, said Tim Hwang, who worked on A.I. policy at Google and is now the director of the Ethics and Governance of AI Initiative, a new organization created by the LinkedIn founder Reid Hoffman and others to fund ethical research in artificial intelligence. China's embrace of A.I. comes at a crucial time in the development of the technology and just as the lead long enjoyed by the United States has started to dwindle. For decades, artificial intelligence was more fiction than science. In the past few years, however, dramatic improvements have prompted some of the biggest companies in Silicon Valley and Detroit -- and China -- to invest billions on everything from self-driving cars to home appliances that can have a conversation with a human.

In previous posts, I introduced Keras for building convolutional neural networks and performing word embedding. The next natural step is to talk about implementing recurrent neural networks in Keras. In a previous tutorial of mine, I gave a very comprehensive introduction to recurrent neural networks and long short term memory (LSTM) networks, implemented in TensorFlow. In this tutorial, I'll concentrate on creating LSTM networks in Keras, briefly giving a recap or overview of how LSTMs work. In this Keras LSTM tutorial, we'll implement a sequence-to-sequence text prediction model by utilizing a large text data set called the PTB corpus. All the code in this tutorial can be found on this site's Github repository. Recommended online course: If you are more of a video course learner, I'd recommend this inexpensive Udemy course to learn more about Keras and LSTM networks: Zero to Deep Learning with Python and Keras A LSTM network is a kind of recurrent neural network.

Your Gmail inbox is about to get weird. As part of an update to its Accelerated Mobile Pages project, Google will serve up content from the internet inside e-mails to provide always-up-to-date information. Backstory: The AMP project was designed to make web pages load faster, so you could click through from search results to content almost instantly. Now Google wants to do some ... interesting things with the technology. E-mail plus plus: Developers are going to be playing around with AMP widgets for Gmail messages.