To understand how artificial intelligence is fundamentally changing how Jumio verifies online IDs and identities, you must first understand two concepts you've likely heard of: machine learning and deep learning. Definitionally, machine learning are algorithms that parse data, learn from that data, and then apply what they've learned to make informed decisions. So rather than hand-coding software routines with a specific set of instructions to accomplish a particular task, the machine is "trained" using large amounts of data and algorithms that give it the ability to learn how to perform the task. They're only capable of what they're designed for; nothing more, nothing less. As it turned out, one of the very best application areas for machine learning was classical computer vision, though it still required a great deal of hand-coding of specific features to get the job done.

Through extensive systematic experiments, we show how these traditional approaches fail to explain why large neural networks generalize well in practice. Specifically, our experiments establish that state-of-the-art convolutional networks for image classification trained with stochastic gradient methods easily fit a random labeling of the training data. This phenomenon is qualitatively unaffected by explicit regularization, and occurs even if we replace the true images by completely unstructured random noise. We corroborate these experimental findings with a theoretical construction showing that simple depth two neural networks already have perfect finite sample expressivity as soon as the number of parameters exceeds the number of data points as it usually does in practice.

It's no wonder Dr. Elliot Fishman sounds frustrated when he talks about pancreatic cancer. As a diagnostic radiologist at Johns Hopkins Hospital, one of the world's largest centers for pancreatic cancer treatment, he has the grim task of examining pancreatic CT scans for signs of a disease that's usually too advanced to treat. Because symptoms seldom show up in the early stages of pancreatic cancer, most patients don't get CT scans or other tests until the cancer has spread. By then, the odds of survival are low: Just 7 percent of patients live five years after diagnosis, the lowest rate for any cancer. "Our goal is early detection of pancreatic cancer, and that would save lives," Fishman said.

With Artificial Intelligence (AI) revolutionizing all walks of business and social life, enterprises are plugging in conversational AI to enhance cross channel customer experience. Gartner predicts that more than 50 percent of enterprises will opt for chatbots as the preferred channel over traditional mobile apps. The technology at the core of the rise of the chatbot is natural language processing (NLP). NLP infused chatbots, designed to mimic human conversations, have gone through the peak of Gartner hype cycle expectations, the trough of disillusionment and are heading towards enlightenment and productivity. With more people using chat to communicate, chatbots have gained a lot of importance and are fast becoming a trend for digitally empowered consumers. NLP provides a way for chatbots to understand and interpret the user in their own language to offer a richer conversational experience.

Artificial Neural Networks (ANN) are replacing traditional Machine Learning models to evolve precise and accurate models. Convolutional Neural Networks (CNN) brings the power of deep learning to computer vision. Some of the recent advancements in computer vision such as Single Shot Multibox Detector (SSD) and Generative Adversarial Networks (GAN) are revolutionizing image processing. For example, using some of these techniques, images and videos that are shot in low light and low resolution can be enhanced to HD quality. The ongoing research in computer vision will become the base for image processing in healthcare, defense, transportation and other domains.

Samsung's SVP, head of global research Larry Heck, who joined the company from Google last November, says the new centers represent an acknowledgement that we're still in the early stages of AI, and there are some grand scientific challenges to be solved so that more people actually start using AI every day: "These challenges are going to need the top AI talent in the world to solve them, so what we recognized is that we need to go to where the talent is," says Heck. "Toronto is a world-leading center for research, specifically in deep learning and computer vision, and similarly are Moscow and Cambridge. These are the first three, but we'll be expanding and opening more." The three new centers join Samsung's existing AI-research labs in Seoul and Mountain View, CA. While it does represent a significant boost in the company's AI investment, don't expect the results to land in your new phone next year. Heck says the work at these centers will be tackling advanced research problems, far out in front to provide a path forward for Bixby and their other AI efforts.

Hewlett Packard Enterprise (HPE) has introduced new initiatives to help enterprises ramp up, optimise and scale the use of artificial intelligence (AI) to improve their demand forecasting and operational efficiency, while increasing sales. "Global tech giants are investing heavily in AI, but the majority of enterprises are struggling both with finding viable AI use cases and with building technology environments that support their AI workloads," said Beena Ammanath, global vice president, Artificial Intelligence, HPE Pointnext. "As a result, the gap between leaders and laggards is widening," she said. HPE Digital Prescriptive Maintenance Services from HPE Pointnext leverage AI to predict when industrial equipment will fail, then suggest and automate the correct action to fix the problem before harm is done, increasing the productivity of the industrial equipment. The solution uses enterprise data including real-time and batch data from Internet of Things (IoT) devices, data centres, and the cloud.

This is a PyTorch implementation of Listen, Attend and Spell (LAS) published in ICASSP 2016 (Student Paper Award). Please feel free to use/modify them, any bug report or improvement suggestion will be appreciated. This implement achieves about 34% phoneme error rate on TIMIT's testing set (using original setting in the paper without hyper parameter tuning, models are stored in checkpoint/). It's not a remarkable score but please notice that deep end2end ASR without special designed loss function such as LAS requires larger corpus to achieve outstanding performance. Result of the first sample in TIMIT testing set.

We propose a temporally coherent generative model addressing the super-resolution problem for fluid flows. Our work represents a first approach to synthesize four-dimensional physics fields with neural networks. Based on a conditional generative adversarial network that is designed for the inference of three-dimensional volumetric data, our model generates consistent and detailed results by using a novel temporal discriminator, in addition to the commonly used spatial one. Our experiments show that the generator is able to infer more realistic high-resolution details by using additional physical quantities, such as low-resolution velocities or vorticities. Besides improvements in the training process and in the generated outputs, these inputs offer means for artistic control as well. We additionally employ a physics-aware data augmentation step, which is crucial to avoid overfitting and to reduce memory requirements. In this way, our network learns to generate advected quantities with highly detailed, realistic, and temporally coherent features. Our method works instantaneously, using only a single time-step of low-resolution fluid data. We demonstrate the abilities of our method using a variety of complex inputs and applications in two and three dimensions.

Inspired by how humans summarize long documents, we propose an accurate and fast summarization model that first selects salient sentences and then rewrites them abstractively (i.e., compresses and paraphrases) to generate a concise overall summary. We use a novel sentence-level policy gradient method to bridge the non-differentiable computation between these two neural networks in a hierarchical way, while maintaining language fluency. Empirically, we achieve the new state-of-the-art on all metrics (including human evaluation) on the CNN/Daily Mail dataset, as well as significantly higher abstractiveness scores. Moreover, by first operating at the sentence-level and then the word-level, we enable parallel decoding of our neural generative model that results in substantially faster (10-20x) inference speed as well as 4x faster training convergence than previous long-paragraph encoder-decoder models. We also demonstrate the generalization of our model on the test-only DUC-2002 dataset, where we achieve higher scores than a state-of-the-art model.