Most deep latent factor models choose simple priors for simplicity, tractability or not knowing what prior to use. Recent studies show that the choice of the prior may have a profound effect on the expressiveness of the model,especially when its generative network has limited capacity. In this paper, we propose to learn a proper prior from data for adversarial autoencoders(AAEs). We introduce the notion of code generators to transform manually selected simple priors into ones that can better characterize the data distribution. Experimental results show that the proposed model can generate better image quality and learn better disentangled representations than AAEs in both supervised and unsupervised settings. Lastly, we present its ability to do cross-domain translation in a text-to-image synthesis task.

We introduce WIQA, the first large-scale dataset of "What if..." questions over procedural text. WIQA contains three parts: a collection of paragraphs each describing a process, e.g., beach erosion; a set of crowdsourced influence graphs for each paragraph, describing how one change a ffects another; and a large (40k) collection of "What if...?" multiple-choice questions derived from the graphs. For example, given a paragraph about beach erosion, would stormy weather result in more or less erosion (or have no e ff ect)? The task is to answer the questions, given their associated paragraph. WIQA contains three kinds of questions: perturbations to steps mentioned in the paragraph; external (out-of-paragraph) perturbations requiring commonsense knowledge; and irrelevant (no e ff ect) perturbations. We find that state-of-the-art models achieve 73.8% accuracy, well below the human performance of 96.3%. We analyze the challenges, in particular tracking chains of influences, and present the dataset as an open challenge to the community.

A core capability of intelligent systems is the ability to quickly learn new tasks by drawing on prior experience. Gradient (or optimization) based meta-learning has recently emerged as an effective approach for few-shot learning. In this formulation, meta-parameters are learned in the outer loop, while task-specific models are learned in the inner-loop, by using only a small amount of data from the current task. A key challenge in scaling these approaches is the need to differentiate through the inner loop learning process, which can impose considerable computational and memory burdens. By drawing upon implicit differentiation, we develop the implicit MAML algorithm, which depends only on the solution to the inner level optimization and not the path taken by the inner loop optimizer. This effectively decouples the meta-gradient computation from the choice of inner loop optimizer. As a result, our approach is agnostic to the choice of inner loop optimizer and can gracefully handle many gradient steps without vanishing gradients or memory constraints. Theoretically, we prove that implicit MAML can compute accurate meta-gradients with a memory footprint that is, up to small constant factors, no more than that which is required to compute a single inner loop gradient and at no overall increase in the total computational cost. Experimentally, we show that these benefits of implicit MAML translate into empirical gains on few-shot image recognition benchmarks.

Nature's spectacular inventiveness, reflected in the enormous diversity of form and function displayed by the biosphere, is a feature of life that distinguishes living most strongly from nonliving. It is, therefore, not surprising that this aspect of life should become a central focus of artificial life. We have known since Darwin that the diversity is produced dynamically, through the process of evolution; this has led life's creative productivity to be called Open-Ended Evolution (OEE) in the field. This article introduces the second of two special issues on current research in OEE and provides an overview of the contents of both special issues. Most of the work was presented at a workshop on open-ended evolution that was held as a part of the 2018 Conference on Artificial Life in Tokyo, and much of it had antecedents in two previous workshops on open-ended evolution at artificial life conferences in Cancun and York. We present a simplified categorization of OEE and summarize progress in the field as represented by the articles in this special issue.

This is the next blog in my random series on better understanding some of these advanced Artificial Intelligence and Deep Learning algorithms. This "episode" takes on Generative Adversarial Networks (GANs). Hope you enjoy my "Deep Learning" learning journey. I originally wrote in "Transforming from Autonomous to Smart: Reinforcement Learning Basics" how Reinforcement Learning was creating learning agents to beat games such as Chess, Go and Mario Bros. Reinforcement learning creates intelligent agents that learn via trial-and-error how to map situations to actions so as to maximize rewards. Reinforcement Learning is one of the more powerful Artificial Intelligence (AI) concepts because it is designed to learn and circumnavigate "situations" where you don't have data sets with explicit known outcomes (which represents most real-life situations, like operating an autonomous vehicle).

PLYMOUTH, Mass., September 9, 2019 -- Microway, a leading provider of computational clusters, servers, and workstations for AI and HPC applications, announces it has provided Oregon State University with six NVIDIA DGX-2 supercomputer systems, deployment services, and bringup expertise. Each DGX-2 packs 16 fully connected Tesla V100 GPUs, giving Oregon State a linked network of the world's most powerful AI systems powered by 96 GPU accelerators. The new, massively increased computing capabilities at the College of Engineering resolved a significant campus hardware gap and helped support cutting-edge research on medical imaging, nuclear science, bridge construction, robotics, and driverless vehicles. When planning expanded capability, university faculty and administrators determined they needed enough GPU capacity to serve the diverse needs of undergraduate classes and research workloads, plus lightning-fast storage. The University selected the NVIDIA DGX-2 platform for its immense power, technical support services, and the Docker images with NVIDIA's NGC containerized software.

The rapid pace of innovation in Artificial Intelligence (AI) is creating enormous opportunity for transforming entire industries and our very existence. After competing this comprehensive 6 course Professional Certificate, you will get a practical understanding of Machine Learning and Deep Learning. You will master fundamental concepts of Machine Learning and Deep Learning, including supervised and unsupervised learning. You will utilize popular Machine Learning and Deep Learning libraries such as SciPy, ScikitLearn, Keras, PyTorch, and Tensorflow applied to industry problems involving object recognition and Computer Vision, image and video processing, text analytics, Natural Language Processing, recommender systems, and other types of classifiers. You will be able to scale Machine Learning on Big Data using Apache Spark.

John C. Mauro is Professor of Materials Science and Engineering at the Pennsylvania State University. John earned a B.S. in Glass Engineering Science (2001), B.A. in Computer Science (2001), and Ph.D. in Glass Science (2006), all from Alfred University. He joined Corning Incorporated in 1999 and served in multiple roles there, including Senior Research Manager of the Glass Research department. Mauro is the inventor or co-inventor of several new glass compositions for Corning, including Corning Gorilla Glass products. Mauro joined the faculty at Penn State in 2017 and is currently a world-recognized expert in fundamental and applied glass science, statistical mechanics, computational and condensed matter physics, thermodynamics, and the topology of disordered networks.

In the next three years, as many as 120 million workers in the world's 12 largest economies may need to be retrained or reskilled as a result of AI and intelligent automation, according to a new IBM Institute for Business Value (IBV) study. In addition, only 41 percent of CEOs surveyed say that they have the people, skills and resources required to execute their business strategies. The study, which includes input from more than 5,670 global executives in 48 countries, points to compounding challenges that require a fundamental shift in how companies meet and manage changing workforce needs throughout all levels of the enterprise. According to the global research, the time it takes to close a skills gap through training has increased by more than 10 times in just four years. In 2014, it took three days on average to close a capability gap through training in the enterprise; in 2018, it took 36 days.

As if manufacturers didn't already have enough on their hands trying to find suitable applicants for their shop floors and R&D departments, the world of artificial intelligence is about to explode onto the scene. And when it does, the scramble for talent will only grow maddeningly tougher. This may sound like trouble, but there's a tremendous upside. According to a newly released study by the MAPI Foundation and the Information Technology and Innovation Foundation (ITIF), not only will AI enable machines to do a lot more--but it will also empower humans to do a lot more as well. That means an upsurge of new kinds of jobs related to developing new AI solutions, leading new AI business strategies and supervising AI implementations.