Of the many vehicles that drove the 120 miles from Seoul to Pyeongchang for the Winter Olympics, one Hyundai stood out. Not because it runs on hydrogen, though that's unusual enough. This Nexo crossover did the cross-country trip all on its own. And just like the athletes, the car was there to show off its skills for the crowds. Along with two just like it, plus a pair of autonomous Genesis G80 sedans, it spent the Games giving demonstration rides to fans.

Neural networks are powerful things, but they need a lot of juice. Engineers at MIT have now developed a new chip that cuts neural nets' power consumption by up to 95 percent, potentially allowing them to run on battery-powered mobile devices. Smartphones these days are getting truly smart, with ever more AI-powered services like digital assistants and real-time translation. But typically the neural nets crunching the data for these services are in the cloud, with data from smartphones ferried back and forth. That's not ideal, as it requires a lot of communication bandwidth and means potentially sensitive data is being transmitted and stored on servers outside the user's control.

The contextual bandit literature has traditionally focused on algorithms that address the exploration-exploitation tradeoff. In particular, greedy algorithms that exploit current estimates without any exploration may be sub-optimal in general. However, exploration-free greedy algorithms are desirable in practical settings where exploration may be costly or unethical (e.g., clinical trials). Surprisingly, we find that a simple greedy algorithm can be rate-optimal if there is sufficient randomness in the observed contexts. We prove that this is always the case for a two-armed bandit under a general class of context distributions that satisfy a condition we term covariate diversity. Furthermore, even absent this condition, we show that a greedy algorithm can be rate-optimal with nonzero probability. Thus, standard bandit algorithms may unnecessarily explore. Motivated by these results, we introduce Greedy-First, a new algorithm that uses only observed contexts and rewards to determine whether to follow a greedy algorithm or to explore. We prove that this algorithm is rate-optimal without any additional assumptions on the context distribution or the number of arms. Extensive simulations demonstrate that Greedy-First successfully reduces experimentation and outperforms existing (exploration-based) contextual bandit algorithms such as Thompson sampling or UCB.

Stochastic gradient descent (SGD) has achieved great success in training deep neural network, where the gradient is computed through back-propagation. However, the back-propagated values of different layers vary dramatically. This inconsistence of gradient magnitude across different layers renders optimization of deep neural network with a single learning rate problematic. We introduce the back-matching propagation which computes the backward values on the layer's parameter and the input by matching backward values on the layer's output. This leads to solving a bunch of least-squares problems, which requires high computational cost. We then reduce the back-matching propagation with approximations and propose an algorithm that turns to be the regular SGD with a layer-wise adaptive learning rate strategy. This allows an easy implementation of our algorithm in current machine learning frameworks equipped with auto-differentiation. We apply our algorithm in training modern deep neural networks and achieve favorable results over SGD.

We tackle here the problem of multimodal image non-rigid registration, which is of prime importance in remote sensing and medical imaging. The difficulties encountered by classical registration approaches include feature design and slow optimization by gradient descent. By analyzing these methods, we note the significance of the notion of scale. We design easy-to-train, fully-convolutional neural networks able to learn scale-specific features. Once chained appropriately, they perform global registration in linear time, getting rid of gradient descent schemes by predicting directly the deformation.We show their performance in terms of quality and speed through various tasks of remote sensing multimodal image alignment. In particular, we are able to register correctly cadastral maps of buildings as well as road polylines onto RGB images, and outperform current keypoint matching methods.

With a rapidly increasing number of companies using some form of artificial intelligence (AI), such as big data automation, predictive/prescriptive analytics, machine learning, expert systems, neural networks, interactive voice response technologies, and avatar technologies, in their business models, artificial intelligence is forecast to disrupt all industries. With only a small percentage of businesses not yet using or not even planning to utilize artificial intelligence in any way, some opinions state that within a decade from now, managers not using AI will be replaced by those who do. The main reasons for applying various forms of AI, as the findings of the study "Amplifying Human Potential: Towards Purposeful Artificial Intelligence" reveal, were: As Franklin Wolfe writes in How Artificial Intelligence Will Revolutionize the Energy Industry, a special edition on Harvard University s blog on August 28, 2017, artificial intelligence and the energy sector are becoming more and more interconnected, whereby choosing a career path in either of these sectors does not necessarily signify excluding the other. Phil Goldstein, on the other hand, writes in his article in BizTech on October 25, 2017 that AI can support the energy industry in many ways: in improving energy efficiency, predicting possible blackouts and failures, and even support human beings in detecting completely new sources of energy. According to technology research and advisory firm Gartner, 85% of all customer interactions will be managed without a human by 2020.

Though distribution system operators have been adding more sensors to their networks, they still often lack an accurate real-time picture of the behavior of distributed energy resources such as demand responsive electric loads and residential solar generation. Such information could improve system reliability, economic efficiency, and environmental impact. Rather than installing additional, costly sensing and communication infrastructure to obtain additional real-time information, it may be possible to use existing sensing capabilities and leverage knowledge about the system to reduce the need for new infrastructure. In this paper, we disaggregate a distribution feeder's demand measurements into: 1) the demand of a population of air conditioners, and 2) the demand of the remaining loads connected to the feeder. We use an online learning algorithm, Dynamic Fixed Share (DFS), that uses the real-time distribution feeder measurements as well as models generated from historical building- and device-level data. We develop two implementations of the algorithm and conduct case studies using real demand data from households and commercial buildings to investigate the effectiveness of the algorithm. The case studies demonstrate that DFS can effectively perform online disaggregation and the choice and construction of models included in the algorithm affects its accuracy, which is comparable to that of a set of Kalman filters.

We consider the problem of \emph{fully decentralized} multi-agent reinforcement learning (MARL), where the agents are located at the nodes of a time-varying communication network. Specifically, we assume that the reward functions of the agents might correspond to different tasks, and are only known to the corresponding agent. Moreover, each agent makes individual decisions based on both the information observed locally and the messages received from its neighbors over the network. Within this setting, the collective goal of the agents is to maximize the globally averaged return over the network through exchanging information with their neighbors. To this end, we propose two decentralized actor-critic algorithms with function approximation, which are applicable to large-scale MARL problems where both the number of states and the number of agents are massively large. Under the decentralized structure, the actor step is performed individually by each agent with no need to infer the policies of others. For the critic step, we propose a consensus update via communication over the network. Our algorithms are fully incremental and can be implemented in an online fashion. Convergence analyses of the algorithms are provided when the value functions are approximated within the class of linear functions. Extensive simulation results with both linear and nonlinear function approximations are presented to validate the proposed algorithms. Our work appears to be the first study of fully decentralized MARL algorithms for networked agents with function approximation, with provable convergence guarantees.

This paper offers a multi-disciplinary review of knowledge acquisition methods in human activity systems. The review captures the degree of involvement of various types of agencies in the knowledge acquisition process, and proposes a classification with three categories of methods: the human agent, the human-inspired agent, and the autonomous machine agent methods. In the first two categories, the acquisition of knowledge is seen as a cognitive task analysis exercise, while in the third category knowledge acquisition is treated as an autonomous knowledge-discovery endeavour. The motivation for this classification stems from the continuous change over time of the structure, meaning and purpose of human activity systems, which are seen as the factor that fuelled researchers' and practitioners' efforts in knowledge acquisition for more than a century. We show through this review that the KA field is increasingly active due to the higher and higher pace of change in human activity, and conclude by discussing the emergence of a fourth category of knowledge acquisition methods, which are based on red-teaming and co-evolution.

Since the late 1980s particle physicists have used AI even as the concept of a neural network was barely in the public's consciousness. AI and particle physics go hand in hand as the experiments the physicists perform usually revolves around seeking out patterns in the data from particle detectors and AI is excellent at pattern detection. Boaz Klima, a Physicists from the Fermi National Accelerator Laboratory, also called Fermilab, says "It took us several years to convince people that this is not just some magic, hocus-pocus, black box stuff." He was amongst the first to adopt AI tools but today, it's a part of standard particle physics practices. Usually, particle physicists aim to comprehend the way the inner gears of the universe works, typically by colliding subatomic particles at hit speeds to break them down into even smaller and more unusual kinds of matter.