Sponsored You've got an application or workflow that needs to do lots of repetitive work typically done by people in the past. Or you want to branch out into some new area of digital business or customer experience. It might just be a perfect fit for an artificial intelligence algorithm, perhaps using a form of machine learning or a rules-based system. AI has been a long-promised concept but has been held back by, among other factors, a lack of the kinds of raw computer power required to process vast amounts of data and to crunch complex algorithms. It's only now, thanks to cloud, these resources are becoming available, as Intel and service providers make available this kind of power available through their massive server farms.

Deep neural networks--a form of artificial intelligence--have demonstrated mastery of tasks once thought uniquely human. Their triumphs have ranged from identifying animals in images, to recognizing human speech, to winning complex strategy games, among other successes. Now, researchers are eager to apply this computational technique--commonly referred to as deep learning--to some of science's most persistent mysteries. But because scientific data often looks much different from the data used for animal photos and speech, developing the right artificial neural network can feel like an impossible guessing game for nonexperts. To expand the benefits of deep learning for science, researchers need new tools to build high-performing neural networks that don't require specialized knowledge.

There are many established and startup companies developing deep learning chips. Google and Wave Computing have working silicon and are conducting customer trials. Chinese AI chip startup has received $100 million in funding. Cambricon Technologies aims to have one billion smart devices using its AI processor and own 30% of China's high-performance AI chip market in three years. Huawei estimates Cambricon chips are six times faster for deep-learning applications like training algorithms to identify images than a GPU.

Dell EMC has announced the launch of its new machine learning and deep learning solutions, which according to the company is in line with it continuing its work to bring high-performance computing (HPC) and data analytics capabilities to mainstream enterprises worldwide. Dell EMC believes that this enables organisations to take advantage of the convergence of HPC and data analytics and realise advancements in areas including fraud detection, image processing, financial investment analysis and personalised medicine. According to the company, these new innovations represent the next step in the company's focus on democratising HPC, optimising data analytics with artificial intelligence (AI) technology innovations, and advancing both the HPC and AI communities. While AI techniques, such as machine learning and deep learning, being rapidly being deployed by many organisations across several industries, only a small number possess the expertise to design, deploy and manage such systems to use them effectively for rapidly gaining new insights. Dell EMC believes that by leveraging Dell's ecosystem of partnerships and internal expertise in HPC and data analytics services, the company's new solutions offer customers the ability to harness the power of the massive amounts of their collected data, delivering faster, better and deeper business insights in real-time.

For years, artificial intelligence and process automation have evolved largely on independent paths. The coming year will bring steady innovation in both fields, but the real leap forward will be the synergy created when AI and intelligent process automation are harnessed together. To explain what I mean, I'll start by spending a few minutes on each area individually. In AI and machine learning, high-value targets like speech recognition and synthesis, image recognition and bots were identified years ago, but have seemed to remain just out of reach. Now these capabilities are becoming widely appreciated due to advances in three critical areas: massive compute power, robust deep-learning algorithms, and the availability of huge volumes of data to train models.

Recent studies have revealed the vulnerability of deep neural networks - A small adversarial perturbation that is imperceptible to human can easily make a well-trained deep neural network mis-classify. This makes it unsafe to apply neural networks in security-critical applications. In this paper, we propose a new defensive algorithm called Random Self-Ensemble (RSE) by combining two important concepts: ${\bf randomness}$ and ${\bf ensemble}$. To protect a targeted model, RSE adds random noise layers to the neural network to prevent from state-of-the-art gradient-based attacks, and ensembles the prediction over random noises to stabilize the performance. We show that our algorithm is equivalent to ensemble an infinite number of noisy models $f_\epsilon$ without any additional memory overhead, and the proposed training procedure based on noisy stochastic gradient descent can ensure the ensemble model has good predictive capability. Our algorithm significantly outperforms previous defense techniques on real datasets. For instance, on CIFAR-10 with VGG network (which has $92\%$ accuracy without any attack), under the state-of-the-art C&W attack within a certain distortion tolerance, the accuracy of unprotected model drops to less than $10\%$, the best previous defense technique has $48\%$ accuracy, while our method still has $86\%$ prediction accuracy under the same level of attack. Finally, our method is simple and easy to integrate into any neural network.

We use a deep learning model trained only on a patient's blood oxygenation data (measurable with an inexpensive fingertip sensor) to predict impending hypoxemia (low blood oxygen) more accurately than trained anesthesiologists with access to all the data recorded in a modern operating room. We also provide a simple way to visualize the reason why a patient's risk is low or high by assigning weight to the patient's past blood oxygen values. This work has the potential to provide cutting-edge clinical decision support in low-resource settings, where rates of surgical complication and death are substantially greater than in high-resource areas.

An intriguing property of deep neural networks is their inherent vulnerability to adversarial inputs, which significantly hinders their application in security-critical domains. Most existing detection methods attempt to use carefully engineered patterns to distinguish adversarial inputs from their genuine counterparts, which however can often be circumvented by adaptive adversaries. In this work, we take a completely different route by leveraging the definition of adversarial inputs: while deceiving for deep neural networks, they are barely discernible for human visions. Building upon recent advances in interpretable models, we construct a new detection framework that contrasts an input's interpretation against its classification. We validate the efficacy of this framework through extensive experiments using benchmark datasets and attacks. We believe that this work opens a new direction for designing adversarial input detection methods.

Deep neural networks are commonly developed and trained in 32-bit floating point format. Significant gains in performance and energy efficiency could be realized by training and inference in numerical formats optimized for deep learning. Despite advances in limited precision inference in recent years, training of neural networks in low bit-width remains a challenging problem. Here we present the Flexpoint data format, aiming at a complete replacement of 32-bit floating point format training and inference, designed to support modern deep network topologies without modifications. Flexpoint tensors have a shared exponent that is dynamically adjusted to minimize overflows and maximize available dynamic range. We validate Flexpoint by training AlexNet, a deep residual network and a generative adversarial network, using a simulator implemented with the neon deep learning framework. We demonstrate that 16-bit Flexpoint closely matches 32-bit floating point in training all three models, without any need for tuning of model hyperparameters. Our results suggest Flexpoint as a promising numerical format for future hardware for training and inference.

Augmenting a neural network with memory that can grow without growing the number of trained parameters is a recent powerful concept with many exciting applications. We propose a design of memory augmented neural networks (MANNs) called Labeled Memory Networks (LMNs) suited for tasks requiring online adaptation in classification models. LMNs organize the memory with classes as the primary key.The memory acts as a second boosted stage following a regular neural network thereby allowing the memory and the primary network to play complementary roles. Unlike existing MANNs that write to memory for every instance and use LRU based memory replacement, LMNs write only for instances with non-zero loss and use label-based memory replacement. We demonstrate significant accuracy gains on various tasks including word-modelling and few-shot learning. In this paper, we establish their potential in online adapting a batch trained neural network to domain-relevant labeled data at deployment time. We show that LMNs are better than other MANNs designed for meta-learning. We also found them to be more accurate and faster than state-of-the-art methods of retuning model parameters for adapting to domain-specific labeled data.