The frequency and intensity of natural disasters has significantly increased over the past decades and this trend is predicted to continue. Facing these possible and unexpected disasters, urban emergency management has become the especially important issue for the whole governments around the world. In this paper, we present a novel intelligent system for urban emergency management during the large-scale disasters. The proposed system stores and manages the global positioning system (GPS) records from mobile devices used by approximately 1.6 million people throughout Japan over one year. By mining and analyzing population movements after the Great East Japan Earthquake, our system can automatically learn a probabilistic model to better understand and simulate human mobility during the emergency situations. Based on the learning model, population mobility in various urban areas impacted by the earthquake throughout Japan can be automatically simulated or predicted. On the basis of such kind of system, it is easy for us to find some new features or population mobility patterns after the recent and unprecedented composite disasters, which are likely to provide valuable experience and play a vital role for future disaster management worldwide. Figure 1: What kinds of experiences or model can we learn from the unprecedented composite disaster of Japan in 2011?

Researchers have introduced the Dynamic Distributed Constraint Optimization Problem (Dynamic DCOP) formulation to model dynamically changing multi-agent coordination problems, where a dynamic DCOP is a sequence of (static canonical) DCOPs, each partially different from the DCOP preceding it. Existing work typically assumes that the problem in each time step is decoupled from the problems in other time steps, which might not hold in some applications. Therefore, in this paper, we make the following contributions: (i) We introduce a new model, called Markovian Dynamic DCOPs (MD-DCOPs), where the DCOP in the next time step is a function of the value assignments in the current time step; (ii) We introduce two distributed reinforcement learning algorithms, the Distributed RVI Q-learning algorithm and the Distributed R-learning algorithm, that balance exploration and exploitation to solve MD-DCOPs in an online manner; and (iii) We empirically evaluate them against an existing multi-arm bandit DCOP algorithm on dynamic DCOPs.

The growing Electric Vehicles' (EVs) popularity among commuters creates new challenges for the smart grid. The most important of them is the uncoordinated EV charging that substantially increases the energy demand peaks, putting the smart grid under constant strain. In order to cope with these peaks the grid needs extra infrastructure, a costly solution. We propose an Adaptive Management of EV Storage (AMEVS) algorithm, implemented through a learning agent that acts on behalf of individual EV owners and schedules EV charging over a weekly horizon. It accounts for individual preferences so that mobility service is not violated but also individual benefit is maximized. We observe that it reshapes the energy demand making it less volatile so that fewer resources are needed to cover peaks. It assumes Vehicle-to-Grid discharging when the customer has excess capacity. Our agent uses Reinforcement Learning trained on real world data to learn individual household consumption behavior and to schedule EV charging. Unlike previous work, AMEVS is a fully distributed approach. We show that AMEVS achieves significant reshaping of the energy demand curve and peak reduction, which is correlated with customer preferences regarding perceived utility of energy availability. Additionally, we show that the average and peak energy prices are reduced as a result of smarter energy use.

This paper introduces two novel algorithms for learning behaviors from human-provided rewards. The primary novelty of these algorithms is that instead of treating the feedback as a numeric reward signal, they interpret feedback as a form of discrete communication that depends on both the behavior the trainer is trying to teach and the teaching strategy used by the trainer. For example, some human trainers use a lack of feedback to indicate whether actions are correct or incorrect, and interpreting this lack of feedback accurately can significantly improve learning speed. Results from user studies show that humans use a variety of training strategies in practice and both algorithms can learn a contextual bandit task faster than algorithms that treat the feedback as numeric. Simulated trainers are also employed to evaluate the algorithms in both contextual bandit and sequential decision-making tasks with similar results.

Approximate dynamic programming (ADP) has proven itself in a wide range of applications spanning large-scale transportation problems, health care, revenue management, and energy systems. The design of effective ADP algorithms has many dimensions, but one crucial factor is the stepsize rule used to update a value function approximation. Many operations research applications are computationally intensive, and it is important to obtain good results quickly. Furthermore, the most popular stepsize formulas use tunable parameters and can produce very poor results if tuned improperly. We derive a new stepsize rule that optimizes the prediction error in order to improve the short-term performance of an ADP algorithm. With only one, relatively insensitive tunable parameter, the new rule adapts to the level of noise in the problem and produces faster convergence in numerical experiments.

Many problems in computational sustainability require making a sequence of decisions in complex, uncertain environments. In this article, we review the recently discovered notion of adaptive submodularity, an intuitive diminishing returns condition that generalizes the classical notion of submodular set functions to sequential decision problems. We illustrate this concept in several case studies of interest in computational sustainability: First, we demonstrate how it can be used to efficiently plan for resolving uncertainty in adaptive management scenarios. Secondly, we show how it applies to dynamic conservation planning for protecting endangered species, a case study carried out in collaboration with the US Geological Survey and the US Fish and Wildlife Service.

Tackling large approximate dynamic programming or reinforcement learning problems requires methods that can exploit regularities, or intrinsic structure, of the problem in hand. Most current methods are geared towards exploiting the regularities of either the value function or the policy. We introduce a general classification-based approximate policy iteration (CAPI) framework, which encompasses a large class of algorithms that can exploit regularities of both the value function and the policy space, depending on what is advantageous. This framework has two main components: a generic value function estimator and a classifier that learns a policy based on the estimated value function. We establish theoretical guarantees for the sample complexity of CAPI-style algorithms, which allow the policy evaluation step to be performed by a wide variety of algorithms (including temporal-difference-style methods), and can handle nonparametric representations of policies. Our bounds on the estimation error of the performance loss are tighter than existing results. We also illustrate this approach empirically on several problems, including a large HIV control task.

In both the fields of computer science and medicine there is very strong interest in developing personalized treatment policies for patients who have variable responses to treatments. In particular, I aim to find an optimal personalized treatment policy which is a non-deterministic function of the patient specific covariate data that maximizes the expected survival time or clinical outcome. I developed an algorithmic framework to solve multistage decision problem with a varying number of stages that are subject to censoring in which the "rewards" are expected survival times. In specific, I developed a novel Q-learning algorithm that dynamically adjusts for these parameters. Furthermore, I found finite upper bounds on the generalized error of the treatment paths constructed by this algorithm. I have also shown that when the optimal Q-function is an element of the approximation space, the anticipated survival times for the treatment regime constructed by the algorithm will converge to the optimal treatment path. I demonstrated the performance of the proposed algorithmic framework via simulation studies and through the analysis of chronic depression data and a hypothetical clinical trial. The censored Q-learning algorithm I developed is more effective than the state of the art clinical decision support systems and is able to operate in environments when many covariate parameters may be unobtainable or censored.

Demand response (DR) for residential and small commercial buildings is estimated to account for as much as 65% of the total energy savings potential of DR, and previous work shows that a fully automated Energy Management System (EMS) is a necessary prerequisite to DR in these areas. In this paper, we propose a novel EMS formulation for DR problems in these sectors. Specifically, we formulate a fully automated EMS's rescheduling problem as a reinforcement learning (RL) problem, and argue that this RL problem can be approximately solved by decomposing it over device clusters. Compared with existing formulations, our new formulation (1) does not require explicitly modeling the user's dissatisfaction on job rescheduling, (2) enables the EMS to self-initiate jobs, (3) allows the user to initiate more flexible requests and (4) has a computational complexity linear in the number of devices. We also demonstrate the simulation results of applying Q-learning, one of the most popular and classical RL algorithms, to a representative example.

We study online learning of finite Markov decision process (MDP) problems when a side information vector is available. The problem is motivated by applications such as clinical trials, recommendation systems, etc. Such applications have an episodic structure, where each episode corresponds to a patient/customer. Our objective is to compete with the optimal dynamic policy that can take side information into account. We propose a computationally efficient algorithm and show that its regret is at most $O(\sqrt{T})$, where $T$ is the number of rounds. To best of our knowledge, this is the first regret bound for this setting.