We prove novel algorithmic guarantees for several online problems in the smoothed analysis model. In this model, at each time an adversary chooses an input distribution with density function bounded above by $\tfrac{1}{\sigma}$ times that of the uniform distribution; nature then samples an input from this distribution. Crucially, our results hold for {\em adaptive} adversaries that can choose an input distribution based on the decisions of the algorithm and the realizations of the inputs in the previous time steps. This paper presents a general technique for proving smoothed algorithmic guarantees against adaptive adversaries, in effect reducing the setting of adaptive adversaries to the simpler case of oblivious adversaries. We apply this technique to prove strong smoothed guarantees for three problems: -Online learning: We consider the online prediction problem, where instances are generated from an adaptive sequence of $\sigma$-smooth distributions and the hypothesis class has VC dimension $d$. We bound the regret by $\tilde{O}\big(\sqrt{T d\ln(1/\sigma)} + d\sqrt{\ln(T/\sigma)}\big)$. This answers open questions of [RST11,Hag18]. -Online discrepancy minimization: We consider the online Koml\'os problem, where the input is generated from an adaptive sequence of $\sigma$-smooth and isotropic distributions on the $\ell_2$ unit ball. We bound the $\ell_\infty$ norm of the discrepancy vector by $\tilde{O}\big(\ln^2\!\big( \frac{nT}{\sigma}\big) \big)$. -Dispersion in online optimization: We consider online optimization of piecewise Lipschitz functions where functions with $\ell$ discontinuities are chosen by a smoothed adaptive adversary and show that the resulting sequence is $\big( {\sigma}/{\sqrt{T\ell}}, \tilde O\big(\sqrt{T\ell} \big)\big)$-dispersed. This matches the parameters of [BDV18] for oblivious adversaries, up to log factors.

We study reinforcement learning in an infinite-horizon average-reward setting with linear function approximation, where the transition probability function of the underlying Markov Decision Process (MDP) admits a linear form over a feature mapping of the current state, action, and next state. We propose a new algorithm UCRL2-VTR, which can be seen as an extension of the UCRL2 algorithm with linear function approximation. We show that UCRL2-VTR with Bernstein-type bonus can achieve a regret of $\tilde{O}(d\sqrt{DT})$, where $d$ is the dimension of the feature mapping, $T$ is the horizon, and $\sqrt{D}$ is the diameter of the MDP. We also prove a matching lower bound $\tilde{\Omega}(d\sqrt{DT})$, which suggests that the proposed UCRL2-VTR is minimax optimal up to logarithmic factors. To the best of our knowledge, our algorithm is the first nearly minimax optimal RL algorithm with function approximation in the infinite-horizon average-reward setting.

High-dimensional black-box optimisation remains an important yet notoriously challenging problem. Despite the success of Bayesian optimisation methods on continuous domains, domains that are categorical, or that mix continuous and categorical variables, remain challenging. We propose a novel solution -- we combine local optimisation with a tailored kernel design, effectively handling high-dimensional categorical and mixed search spaces, whilst retaining sample efficiency. We further derive convergence guarantee for the proposed approach. Finally, we demonstrate empirically that our method outperforms the current baselines on a variety of synthetic and real-world tasks in terms of performance, computational costs, or both.

In collaborative privacy preserving planning (CPPP), a group of agents jointly creates a plan to achieve a set of goals while preserving each others' privacy. During planning, agents often reveal the private dependencies between their public actions to other agents, that is, which public action facilitates the preconditions of another public action. Previous work in CPPP does not limit the disclosure of such dependencies. In this paper, we explicitly limit the amount of disclosed dependencies, allowing agents to publish only a part of their private dependencies. We investigate different strategies for deciding which dependencies to publish, and how they affect the ability to find solutions. We evaluate the ability of two solvers -- distribute forward search and centralized planning based on a single-agent projection -- to produce plans under this constraint. Experiments over standard CPPP domains show that the proposed dependency-sharing strategies enable generating plans while sharing only a small fraction of all private dependencies.

In this paper, we revisit the problem of smoothed online learning, in which the online learner suffers both a hitting cost and a switching cost, and target two performance metrics: competitive ratio and dynamic regret with switching cost. To bound the competitive ratio, we assume the hitting cost is known to the learner in each round, and investigate the greedy algorithm which simply minimizes the weighted sum of the hitting cost and the switching cost. Our theoretical analysis shows that the greedy algorithm, although straightforward, is $1+ \frac{2}{\alpha}$-competitive for $\alpha$-polyhedral functions, $1+O(\frac{1}{\lambda})$-competitive for $\lambda$-quadratic growth functions, and $1 + \frac{2}{\sqrt{\lambda}}$-competitive for convex and $\lambda$-quadratic growth functions. To bound the dynamic regret with switching cost, we follow the standard setting of online convex optimization, in which the hitting cost is convex but hidden from the learner before making predictions. We modify Ader, an existing algorithm designed for dynamic regret, slightly to take into account the switching cost when measuring the performance. The proposed algorithm, named as Smoothed Ader, attains an optimal $O(\sqrt{T(1+P_T)})$ bound for dynamic regret with switching cost, where $P_T$ is the path-length of the comparator sequence. Furthermore, if the hitting cost is accessible in the beginning of each round, we obtain a similar guarantee without the bounded gradient condition.

The Clustered Shortest-Path Tree Problem (CluSPT) plays an important role in various types of optimization problems in real-life. Recently, some Multifactorial Evolutionary Algorithm (MFEA) have been introduced to deal with the CluSPT, however these researches still have some shortcomings such as evolution operators only perform on complete graphs, huge resource consumption for finding the solution on large search spaces. To overcome these limitations, this paper describes a MFEA-based approach to solve the CluSPT. The proposed algorithm utilizes Dijkstra's algorithm to construct the spanning trees in clusters while using evolutionary operators for building the spanning tree connecting clusters. This approach takes advantage of both exact and approximate algorithms so it enables the algorithm to function efficiently on complete and sparse graphs alike. Furthermore, evolutionary operators such as individual encoding and decoding methods are also designed with great consideration regarding performance and memory usage. We have included a proof on the repairing method's efficacy in ensuring all solutions are valid. We have conducted tests on various types of Euclidean instances to assess the effectiveness of the proposed algorithm and methods. Experiment results point out the effectiveness of the proposed algorithm existing heuristic algorithms in most of the test cases. The impact of the proposed MFEA was analyzed and a possible influential factor that may be useful for further study was also pointed out.

In this paper, we present the Adaptive EntropyTree Search (ANTS) algorithm. ANTS builds on recent successes of maximum entropy planning while mitigating its arguably major drawback - sensitivity to the temperature setting. We endow ANTS with a mechanism, which adapts the temperature to match a given range of action selection entropy in the nodes of the planning tree. With this mechanism, the ANTS planner enjoys remarkable hyper-parameter robustness, achieves high scores on the Atari benchmark, and is a capable component of a planning-learning loop akin to AlphaZero. We believe that all these features make ANTS a compelling choice for a general planner for complex tasks.

Academic timetabling is the task of scheduling courses to specific times in such a way that there are no conflicts. Most of the models considered in the literature assume that this conflict information is already known. However in many real life timetabling problems, courses are taught in multiple sections and until a student is assigned to a specific section of a course, the conflict information is not known. M.W. Carter [Car00] sums it up nicely "When courses are offered in multiple sections as they are at Waterloo, it creates a timetabling paradox. Students request a course, but timetabling assigns days and times to course sections. We cannot assign times to sections until we know which students are in each section. But we cannot assign students to sections until we know when the sections are timetabled!"

The family of methods collectively known as classifier chains has become a popular approach to multi-label learning problems. This approach involves chaining together off-the-shelf binary classifiers in a directed structure, such that individual label predictions become features for other classifiers. Such methods have proved flexible and effective and have obtained state-of-the-art empirical performance across many datasets and multi-label evaluation metrics. This performance led to further studies of the underlying mechanism and efficacy, and investigation into how it could be improved. In the recent decade, numerous studies have explored the theoretical underpinnings of classifier chains, and many improvements have been made to the training and inference procedures, such that this method remains among the best options for multi-label learning. Given this past and ongoing interest, which covers a broad range of applications and research themes, the goal of this work is to provide a review of classifier chains, a survey of the techniques and extensions provided in the literature, as well as perspectives for this approach in the domain of multi-label classification in the future. We conclude positively, with a number of recommendations for researchers and practitioners, as well as outlining key issues for future research.

The construction of approximate replication strategies for derivative contracts in incomplete markets is a key problem of financial engineering. Recently Reinforcement Learning algorithms for pricing and hedging under realistic market conditions have attracted significant interest. While financial research mostly focused on variations of Q-learning, in Artificial Intelligence Monte Carlo Tree Search is the recognized stateof-the-art method for various planning problems, such as the games of Hex, Chess, Go,... This article introduces Monte Carlo Tree Search as a method to solve the stochastic optimal control problem underlying the pricing and hedging of financial derivatives. As compared to Q-learning it combines reinforcement learning with tree search techniques. As a consequence Monte Carlo Tree Search has higher sample efficiency, is less prone to over-fitting to specific market models and generally learns stronger policies faster. In our experiments we find that Monte Carlo Tree Search, being the world-champion in games like Chess and Go, is easily capable of directly maximizing the utility of investor's terminal wealth without an intermediate mathematical theory.