Modern learning models are characterized by large hyperparameter spaces. In order to adequately explore these large spaces, we must evaluate a large number of configurations, typically orders of magnitude more configurations than available parallel workers. Given the growing costs of model training, we would ideally like to perform this search in roughly the same wall-clock time needed to train a single model. In this work, we tackle this challenge by introducing ASHA, a simple and robust hyperparameter tuning algorithm with solid theoretical underpinnings that exploits parallelism and aggressive early-stopping. Our extensive empirical results show that ASHA slightly outperforms Fabolas and Population Based Tuning, state-of-the hyperparameter tuning methods; scales linearly with the number of workers in distributed settings; converges to a high quality configuration in half the time taken by Vizier (Google's internal hyperparameter tuning service) in an experiment with 500 workers; and beats the published result for a near state-of-the-art LSTM architecture in under 2x the time to train a single model.

Convolutional neural network (CNN) has been widely used for image processing tasks.In this paper we design a bottleneck supervised U-Net model and apply it to liver and tumor segmentation. Taking an image as input, the model outputs segmented images of the same size, each pixel of which takes value from 1 to K where K is the number of classes to be segmented. The innovations of this paper are two-fold: first we design a novel U-Net structure which include dense block and inception block as the base U-Net; second we design a double U-Net architecture based on the base U-Net and includes an encoding U-Net and a segmentation U-Net. The encoding U-Net is first trained to encode the labels, then the encodings are used to supervise the bottleneck of the segmentation U-Net. While training the segmentation U-Net, a weighted average of dice loss(for the final output) and MSE loss(for the bottleneck) is used as the overall loss function. This approach can help retain the hidden features of input images. The model is applied to a liver tumor 3D CT scan dataset to conduct liver and tumor segmentation sequentially. Experimental results indicate bottleneck supervised U-Net can accomplish segmentation tasks effectively with better performance in controlling shape distortion, reducing false positive and false negative, besides accelerating convergence. Besides, this model has good generalization for further improvement.

There has been a recent explosion in the capabilities of game-playing artificial intelligence. Many classes of tasks, from video games to motor control to board games, are now solvable by fairly generic algorithms, based on deep learning and reinforcement learning, that learn to play from experience with minimal prior knowledge. However, these machines often do not win through intelligence alone -- they possess vastly superior speed and precision, allowing them to act in ways a human never could. To level the playing field, we restrict the machine's reaction time to a human level, and find that standard deep reinforcement learning methods quickly drop in performance. We propose a solution to the action delay problem inspired by human perception -- to endow agents with a neural predictive model of the environment which "undoes" the delay inherent in their environment -- and demonstrate its efficacy against professional players in Super Smash Bros. Melee, a popular console fighting game.

A general-purpose intelligent robot must be able to learn autonomously and be able to accomplish multiple tasks in order to be deployed in the real world. However, standard reinforcement learning approaches learn separate task-specific policies and assume the reward function for each task is known a priori. We propose a framework that learns event cues from off-policy data, and can flexibly combine these event cues at test time to accomplish different tasks. These event cue labels are not assumed to be known a priori, but are instead labeled using learned models, such as computer vision detectors, and then `backed up' in time using an action-conditioned predictive model. We show that a simulated robotic car and a real-world RC car can gather data and train fully autonomously without any human-provided labels beyond those needed to train the detectors, and then at test-time be able to accomplish a variety of different tasks. Videos of the experiments and code can be found at https://github.com/gkahn13/CAPs

Abstract--Applying Machine Learning (ML) to business applications for automation usually faces difficulties when integrating diverse ML dependencies and services, mainly because of the lack of a common ML framework. In most cases, the ML models are developed for applications which are targeted for specific business domain use cases, leading to duplicated effort, and making reuse impossible. This paper presents Acumos, an open platform capable of packaging ML models into portable containerized microservices which can be easily shared via the platform's catalog, and can be integrated into various business applications. We present a case study of packaging sentiment analysis and classification ML models via the Acumos platform, permitting easy sharing with others. We demonstrate that the Acumos platform reduces the technical burden on application developers when applying machine learning models to their business applications. Furthermore, the platform allows the reuse of readily available ML microservices in various business domains. In recent years, there has been tremendous excitement around and interest in the potential of ML technologies.

In this paper we propose to view the acceptance rate of the Metropolis-Hastings algorithm as a universal objective for learning to sample from target distribution - given either as a set of samples or in the form of unnormalized density. To reveal the connection we derive the lower bound on the acceptance rate and treat it as the objective for learning explicit and implicit samplers. The form of the lower bound allows for doubly stochastic gradient optimization in case the target distribution factorizes (i.e. over data points). Bayesian framework and deep learning have become more and more interrelated during recent years. Recently Bayesian deep neural networks were used for estimating uncertainty (Gal & Ghahramani, 2016), ensembling (Gal & Ghahramani, 2016) and model compression (Molchanov et al., 2017). On the other hand, deep neural networks may be used to improve approximate inference in Bayesian models (Kingma & Welling, 2014). Learning modern Bayesian neural networks requires inference in the spaces with dimension up to several million by conditioning the weights of DNN on hundreds of thousands of objects.

Traditional data quality control methods are based on users experience or previously established business rules, and this limits performance in addition to being a very time consuming process with lower than desirable accuracy. Utilizing deep learning, we can leverage computing resources and advanced techniques to overcome these challenges and provide greater value to users. In this paper, we, the authors, first review relevant works and discuss machine learning techniques, tools, and statistical quality models. Second, we offer a creative data quality framework based on deep learning and statistical model algorithm for identifying data quality. Third, we use data involving salary levels from an open dataset published by the state of Arkansas to demonstrate how to identify outlier data and how to improve data quality via deep learning. Finally, we discuss future work.

We propose a novel adversarial multi-task learning scheme, aiming at actively curtailing the inter-talker feature variability while maximizing its senone discriminability so as to enhance the performance of a deep neural network (DNN) based ASR system. We call the scheme speaker-invariant training (SIT). In SIT, a DNN acoustic model and a speaker classifier network are jointly optimized to minimize the senone (tied triphone state) classification loss, and simultaneously mini-maximize the speaker classification loss. A speaker-invariant and senone-discriminative deep feature is learned through this adversarial multi-task learning. With SIT, a canonical DNN acoustic model with significantly reduced variance in its output probabilities is learned with no explicit speaker-independent (SI) transformations or speaker-specific representations used in training or testing. Evaluated on the CHiME-3 dataset, the SIT achieves 4.99% relative word error rate (WER) improvement over the conventional SI acoustic model. With additional unsupervised speaker adaptation, the speaker-adapted (SA) SIT model achieves 4.86% relative WER gain over the SA SI acoustic model.

Clothes shopping is a taxing experience. My eyes get bombarded with too much information. Sales, coupons, colors, toddlers, flashing lights, and crowded aisles are just a few examples of all the signals forwarded to my visual cortex, whether or not I actively try to pay attention. The visual system absorbs an abundance of information. Should I go for that H&M khaki pants?

Making sure artificial intelligence does what we want and behaves in predictable ways will be crucial as the technology becomes increasingly ubiquitous. It's an area frequently neglected in the race to develop products, but DeepMind has now outlined its research agenda to tackle the problem. AI safety, as the field is known, has been gaining prominence in recent years. But it's also recognition of the fact that AI technology is quickly pervading all aspects of our lives, making decisions on everything from what movies we watch to whether we get a mortgage. That's why DeepMind hired a bevy of researchers who specialize in foreseeing the unforeseen consequences of the way we built AI back in 2016.