This paper offers a new hybrid probably asymptotically correct (PAC) reinforcement learning (RL) algorithm for Markov decision processes (MDPs) that intelligently maintains favorable features of its parents. The designed algorithm, referred to as the Dyna-Delayed Q-learning (DDQ) algorithm, combines model-free and model-based learning approaches while outperforming both in most cases. The paper includes a PAC analysis of the DDQ algorithm and a derivation of its sample complexity. Numerical results that support the claim regarding the new algorithm's sample efficiency compared to its parents are showcased in a small grid-world example.

Optimizing a partially observable Markov decision process (POMDP) policy is challenging. The policy graph improvement (PGI) algorithm for POMDPs represents the policy as a fixed size policy graph and improves the policy monotonically. Due to the fixed policy size, computation time for each improvement iteration is known in advance. Moreover, the method allows for compact understandable policies. This report describes the technical details of the PGI [1] and particle based PGI [2] algorithms for POMDPs in a more accessible way than [1] or [2] allowing practitioners and students to understand and implement the algorithms.

A geo-marketplace allows users to be paid for their location data. Users concerned about privacy may want to charge more for data that pinpoints their location accurately, but may charge less for data that is more vague. A buyer would prefer to minimize data costs, but may have to spend more to get the necessary level of accuracy. We call this interplay between privacy, utility, and price \emph{spatial privacy pricing}. We formalize the issues mathematically with an example problem of a buyer deciding whether or not to open a restaurant by purchasing location data to determine if the potential number of customers is sufficient to open. The problem is expressed as a sequential decision making problem, where the buyer first makes a series of decisions about which data to buy and concludes with a decision about opening the restaurant or not. We present two algorithms to solve this problem, including experiments that show they perform better than baselines.

To drive purchase in online advertising, it is of the advertiser's great interest to optimize the sequential advertising strategy whose performance and interpretability are both important. The lack of interpretability in existing deep reinforcement learning methods makes it not easy to understand, diagnose and further optimize the strategy. In this paper, we propose our Deep Intents Sequential Advertising (DISA) method to address these issues. The key part of interpretability is to understand a consumer's purchase intent which is, however, unobservable (called hidden states). In this paper, we model this intention as a latent variable and formulate the problem as a Partially Observable Markov Decision Process (POMDP) where the underlying intents are inferred based on the observable behaviors. Large-scale industrial offline and online experiments demonstrate our method's superior performance over several baselines. The inferred hidden states are analyzed, and the results prove the rationality of our inference.

Many offline unsupervised change point detection algorithms rely on minimizing a penalized sum of segment-wise costs. We extend this framework by proposing to minimize a sum of discrepancies between segments. In particular, we propose to select the change points so as to maximize the cross-entropy between successive segments, balanced by a penalty for introducing new change points. We propose a dynamic programming algorithm to solve this problem and analyze its complexity. Experiments on two challenging datasets demonstrate the advantages of our method compared to three state-of-the-art approaches.

We investigate the sample efficiency of reinforcement learning in a $\gamma$-discounted infinite-horizon Markov decision process (MDP) with state space $\mathcal{S}$ and action space $\mathcal{A}$, assuming access to a generative model. Despite a number of prior work tackling this problem, a complete picture of the trade-offs between sample complexity and statistical accuracy is yet to be determined. In particular, prior results suffer from a sample size barrier, in the sense that their claimed statistical guarantees hold only when the sample size exceeds at least $\frac{|\mathcal{S}||\mathcal{A}|}{(1-\gamma)^2}$ (up to some log factor). The current paper overcomes this barrier by certifying the minimax optimality of model-based reinforcement learning as soon as the sample size exceeds the order of $\frac{|\mathcal{S}||\mathcal{A}|}{1-\gamma}$ (modulo some log factor). More specifically, a perturbed model-based planning algorithm provably finds an $\varepsilon$-optimal policy with an order of $\frac{|\mathcal{S}||\mathcal{A}| }{(1-\gamma)^3\varepsilon^2}\log\frac{|\mathcal{S}||\mathcal{A}|}{(1-\gamma)\varepsilon}$ samples for any $\varepsilon \in (0, \frac{1}{1-\gamma}]$. Along the way, we derive improved (instance-dependent) guarantees for model-based policy evaluation. To the best of our knowledge, this work provides the first minimax-optimal guarantee in a generative model that accommodates the entire range of sample sizes (beyond which finding a meaningful policy is information theoretically impossible).

Abstract: This paper surveys the state-of-the-art in machine ethics, that is, considerations of how to implement ethical behaviour in robots, unmanned autonomous vehicles, or software systems. The emphasis is on covering the breadth of ethical theories being considered by implementors, as well as the implementation techniques being used. There is no consensus on which ethical theory is best suited for any particular domain, nor is there any agreement on which technique is best placed to implement a particular theory. Another unresolved problem in these implementations of ethical theories is how to objectively validate the implementations. The paper discusses the dilemmas being used as validating'whetstones' and whether any alternative validation mechanism exists. Finally, it speculates that an intermediate step of creating domain-specific ethics might be a possible stepping stone towards creating machines that exhibit ethical behaviour. Computers are increasingly a part of the socio-technical systems around us. Domains such as smartgrids, cloud computing, healthcare, and transport are but some examples where computers are deeply embedded. The speed and complexity of decision-making in these domains have meant that humans are ceding more and more autonomy to these computers (Nallur & Clarke 2018). Autonomy, in machines, can be defined as the effective decision-making power over goals, that influences some action in the real-world. For instance, smart traffic lights can autonomically change their timings, depending on the flow and density of traffic on the roads.

Emergency response to incidents such as accidents, medical calls, and fires is one of the most pressing problems faced by communities across the globe. In the last fifty years, researchers have developed statistical, analytical, and algorithmic approaches for designing emergency response management (ERM) systems. In this survey, we present models for incident prediction, resource allocation, and dispatch for emergency incidents. We highlight the strengths and weaknesses of prior work in this domain and explore the similarities and differences between different modeling paradigms. Finally, we present future research directions. To the best of our knowledge, our work is the first comprehensive survey that explores the entirety of ERM systems.

In this thesis, we focus on video action understanding problems from an online and real-time processing point of view. We start with the conversion of the traditional offline spatiotemporal action detection pipeline into an online spatiotemporal action tube detection system. An action tube is a set of bounding connected over time, which bounds an action instance in space and time. Next, we explore the future prediction capabilities of such detection methods by extending the an existing action tube into the future by regression. Later, we seek to establish that online/causal representations can achieve similar performance to that of offline three dimensional (3D) convolutional neural networks (CNNs) on various tasks, including action recognition, temporal action segmentation and early prediction.

Discrete event sequences are ubiquitous, such as an ordered event series of process interactions in Information and Communication Technology systems. Recent years have witnessed increasing efforts in detecting anomalies with discrete-event sequences. However, it still remains an extremely difficult task due to several intrinsic challenges including data imbalance issues, the discrete property of the events, and sequential nature of the data. To address these challenges, in this paper, we propose OC4Seq, a multi-scale one-class recurrent neural network for detecting anomalies in discrete event sequences. Specifically, OC4Seq integrates the anomaly detection objective with recurrent neural networks (RNNs) to embed the discrete event sequences into latent spaces, where anomalies can be easily detected. In addition, given that an anomalous sequence could be caused by either individual events, subsequences of events, or the whole sequence, we design a multi-scale RNN framework to capture different levels of sequential patterns simultaneously. Experimental results on three benchmark datasets show that OC4Seq consistently outperforms various representative baselines by a large margin. Moreover, through both quantitative and qualitative analysis, the importance of capturing multi-scale sequential patterns for event anomaly detection is verified.