Abstract--The past several years have seen both an explosion in the use of Convolutional Neural Networks (CNNs) and the design of accelerators to make CNN inference practical. In the architecture community, the lion share of effort has targeted CNN inference for image recognition. The closely related problem of video recognition has received far less attention as an accelerator target. This is surprising, as video recognition is more computationally intensive than image recognition, and video traffic is predicted to be the majority of internet traffic in the coming years. This paper fills the gap between algorithmic and hardware advances for video recognition by providing a design space exploration and flexible architecture for accelerating 3D Convolutional Neural Networks (3D CNNs)--the core kernel in modern video understanding. When compared to (2D) CNNs used for image recognition, efficiently accelerating 3D CNNs poses a significant engineering challenge due to their large (and variable over time) memory footprint and higher dimensionality. To address these challenges, we design a novel accelerator, called Morph, that can adaptively support different spatial and temporal tiling strategies depending on the needs of each layer of each target 3D CNN. We codesign a software infrastructure alongside the Morph hardware to find good-fit parameters to control the hardware. Evaluated on state-of-the-art 3D CNNs, Morph achieves up to 3.4 (2.5 average) reduction in energy consumption and improves performance/watt by up to 5.1 (4 average) compared to a baseline 3D CNN accelerator, with an area overhead of 5%. Morph further achieves a 15.9 average energy reduction on 3D CNNs when compared to Eyeriss. The rise of Convolutional Neural Networks (CNNs) [1], [2], [3], [4] has marked tremendous progress in image recognition, advancing the state-of-the-art in tasks ranging from handwritten digit [5] to complex object recognition [6], [7]. At their core, CNNs are compute intensive, parallel dot product operations. Combined with their importance, this computation style has made CNNs a natural target for hardware ASIC acceleration, and a rich line of work has made large strides in this direction [8], [9], [10], [11], [12], [13]. Given the recent progress in image recognition, a natural question is whether similar strides have been made for the related problem of video recognition. This work was partially supported by NSF award CCF-1725734 and a DARPA SDH contract. Authors contributed equally to this work. Current state-of-the-art results are achieved using 3-dimensional (3D) CNNs, which generalize (2D) CNNs used for image recognition to account for the time dimension, thereby allowing the model to capture spatiotemporal features.

There is a lot going on in oil markets these days. The Saudis are threatening anyone who dares to question their abuse of human rights with a curtailment of oil production. Meanwhile, a lack of take-away capacity combined with temporary U.S. refinery turnarounds have resulted in a disastrous pricing scenario for Canadian crude oil -- across all grades from heavy to light. We believe during times like these it's important to not to get caught-up in the micro and instead focus on the macro. The fact of the matter is that despite all of the press about new innovation and disruption, global fundamentals in the oil sector really haven't changed that much, with large price responses to marginal changes in supply -- both locally and abroad.

Every two seconds, sensors measuring the United States' electrical grid collect 3 petabytes of data – the equivalent of 3 million gigabytes. Data analysis on that scale is a challenge when crucial information is stored in an inaccessible database. But researchers at Purdue University are working on a solution, combining quantum algorithms with classical computing on small-scale quantum computers to speed up database accessibility. They are using data from the U.S. Department of Energy National Labs' sensors, called phasor measurement units, that collect information on the electrical power grid about voltages, currents and power generation. Because these values can vary, keeping the power grid stable involves continuously monitoring the sensors.

Business spending on cognitive systems will jump an estimated 54% in 2018. Companies continue to invest in artificial intelligence (AI). Doing so means they could boost revenues by 38% within five years. This would also raise employment by 10 percent. By 2030, AI could bump up the global GDP by $15.7 trillion, or a 14% increase.

We discuss deep reinforcement learning in an overview style. We draw a big picture, filled with details. We discuss six core elements, six important mechanisms, and twelve applications, focusing on contemporary work, and in historical contexts. We start with background of artificial intelligence, machine learning, deep learning, and reinforcement learning (RL), with resources. Next we discuss RL core elements, including value function, policy, reward, model, exploration vs. exploitation, and representation. Then we discuss important mechanisms for RL, including attention and memory, unsupervised learning, hierarchical RL, multi-agent RL, relational RL, and learning to learn. After that, we discuss RL applications, including games, robotics, natural language processing (NLP), computer vision, finance, business management, healthcare, education, energy, transportation, computer systems, and, science, engineering, and art. Finally we summarize briefly, discuss challenges and opportunities, and close with an epilogue.

We consider the problem of balancing exploration and exploitation in sequential decision making problems. To explore efficiently, it is vital to consider the uncertainty over all consequences of a decision, and not just those that follow immediately; the uncertainties involved need to be propagated according to the dynamics of the problem. To this end, we develop Successor Uncertainties, a probabilistic model for the state-action value function of a Markov Decision Process that propagates uncertainties in a coherent and scalable way. We relate our approach to other classical and contemporary methods for exploration and present an empirical analysis.

We describe physical tests validating progress made toward acceleration and automation of hydrodynamic codes in the regime of developed turbulence by two {\bf Deep Learning} (DL) Neural Network (NN) schemes trained on {\bf Direct Numerical Simulations} of turbulence. Even the bare DL solutions, which do not take into account any physics of turbulence explicitly, are impressively good overall when it comes to qualitative description of important features of turbulence. However, the early tests have also uncovered some caveats of the DL approaches. We observe that the static DL scheme, implementing Convolutional GAN and trained on spatial snapshots of turbulence, fails to reproduce intermittency of turbulent fluctuations at small scales and details of the turbulence geometry at large scales. We show that the dynamic NN scheme, LAT-NET, trained on a temporal sequence of turbulence snapshots is capable to correct for the small-scale caveat of the static NN. We suggest a path forward towards improving reproducibility of the large-scale geometry of turbulence with NN.

The late physicist Stephen Hawking's last writings predict that a breed of superhumans will take over, having used genetic engineering to surpass their fellow beings. In Brief Answers to the Big Questions, to be published on Oct. 16 and excerpted today in the UK's Sunday Times (paywall), Hawking pulls no punches on subjects like machines taking over, the biggest threat to earth, and the possibilities of intelligent life in space. Hawking delivers a grave warning on the importance of regulating AI, noting that "in the future AI could develop a will of its own, a will that is in conflict with ours." A possible arms race over autonomous-weapons should be stopped before it can start, he writes, asking what would happen if a crash similar to the 2010 stock market Flash Crash happened with weapons. In short, the advent of super-intelligent AI would be either the best or the worst thing ever to happen to humanity.

Technology giant Huawei aims to help train one million artificial intelligence (AI) talents in the next three years to boost the fast-expanding sector. Huawei will provide free online training, organise boot camps and collaborate with industry players. It will also set up a one billion yuan (S$199 million) fund for universities and research institutes to support AI talent development. Mr Zheng Yelai, Huawei's vice-president and president of its cloud business unit, announced this yesterday, the last day of the Huawei Connect Conference in Shanghai. The move is in line with China's push to become a global AI powerhouse in the next decade.

Futurists may delight in predicting a time when business decisions will be made entirely by intelligent algorithms, but today's innovative business executives are making nearer-term arrangements. They are finding ways to use big data analytics and artificial intelligence (AI) to help their employees address two persistent business challenges: providing faster service and making better sales decisions. In both cases, the new approach combines the best of machine learning with the best of human insight to produce better business outcomes. The power of that approach will be felt across a wide swath of the economy, including manufacturers, retailers, banks, insurance companies, real estate firms, utilities, transportation, and the oil and gas industry. Indeed, nearly any business where customer interaction or equipment service is key to success can benefit from this fast-developing concept.