This work revisits the joint beamforming (BF) and antenna selection (AS) problem, as well as its robust beamforming (RBF) version under imperfect channel state information (CSI). Such problems arise due to various reasons, e.g., the costly nature of the radio frequency (RF) chains and energy/resource-saving considerations. The joint (R)BF\&AS problem is a mixed integer and nonlinear program, and thus finding {\it optimal solutions} is often costly, if not outright impossible. The vast majority of the prior works tackled these problems using techniques such as continuous approximations, greedy methods, and supervised machine learning -- yet these approaches do not ensure optimality or even feasibility of the solutions. The main contribution of this work is threefold. First, an effective {\it branch and bound} (B\&B) framework for solving the problems of interest is proposed. Leveraging existing BF and RBF solvers, it is shown that the B\&B framework guarantees global optimality of the considered problems. Second, to expedite the potentially costly B\&B algorithm, a machine learning (ML)-based scheme is proposed to help skip intermediate states of the B\&B search tree. The learning model features a {\it graph neural network} (GNN)-based design that is resilient to a commonly encountered challenge in wireless communications, namely, the change of problem size (e.g., the number of users) across the training and test stages. Third, comprehensive performance characterizations are presented, showing that the GNN-based method retains the global optimality of B\&B with provably reduced complexity, under reasonable conditions. Numerical simulations also show that the ML-based acceleration can often achieve an order-of-magnitude speedup relative to B\&B.

Machine learning (ML) compilers are an active area of research because they offer the potential to automatically speedup tensor programs. Kernel fusion is often cited as an important optimization performed by ML compilers. However, there exists a knowledge gap about how XLA, the most common ML compiler, applies this nuanced optimization, what kind of speedup it can afford, and what low-level effects it has on hardware. Our paper aims to bridge this knowledge gap by studying key compiler passes of XLA's source code. Our evaluation on a reinforcement learning environment Cartpole shows how different fusion decisions in XLA are made in practice. Furthermore, we implement several XLA kernel fusion strategies that can achieve up to 10.56x speedup compared to our baseline implementation.

Household robots operate in the same space for years. Such robots incrementally build dynamic maps that can be used for tasks requiring remote object localization. However, benchmarks in robot learning often test generalization through inference on tasks in unobserved environments. In an observed environment, locating an object is reduced to choosing from among all object proposals in the environment, which may number in the 100,000s. Armed with this intuition, using only a generic vision-language scoring model with minor modifications for 3d encoding and operating in an embodied environment, we demonstrate an absolute performance gain of 9.84% on remote object grounding above state of the art models for REVERIE and of 5.04% on FAO. When allowed to pre-explore an environment, we also exceed the previous state of the art pre-exploration method on REVERIE. Additionally, we demonstrate our model on a real-world TurtleBot platform, highlighting the simplicity and usefulness of the approach. Our analysis outlines a "bag of tricks" essential for accomplishing this task, from utilizing 3d coordinates and context, to generalizing vision-language models to large 3d search spaces.

We study reinforcement learning with linear function approximation where the transition probability and reward functions are linear with respect to a feature mapping $\boldsymbol{\phi}(s,a)$. Specifically, we consider the episodic inhomogeneous linear Markov Decision Process (MDP), and propose a novel computation-efficient algorithm, LSVI-UCB$^+$, which achieves an $\widetilde{O}(Hd\sqrt{T})$ regret bound where $H$ is the episode length, $d$ is the feature dimension, and $T$ is the number of steps. LSVI-UCB$^+$ builds on weighted ridge regression and upper confidence value iteration with a Bernstein-type exploration bonus. Our statistical results are obtained with novel analytical tools, including a new Bernstein self-normalized bound with conservatism on elliptical potentials, and refined analysis of the correction term. This is a minimax optimal algorithm for linear MDPs up to logarithmic factors, which closes the $\sqrt{Hd}$ gap between the upper bound of $\widetilde{O}(\sqrt{H^3d^3T})$ in (Jin et al., 2020) and lower bound of $\Omega(Hd\sqrt{T})$ for linear MDPs.

This work describes in detail how to learn and solve the Rubik's cube game (or puzzle) in the General Board Game (GBG) learning and playing framework. We cover the cube sizes 2x2x2 and 3x3x3. We describe in detail the cube's state representation, how to transform it with twists, whole-cube rotations and color transformations and explain the use of symmetries in Rubik's cube. Next, we discuss different n-tuple representations for the cube, how we train the agents by reinforcement learning and how we improve the trained agents during evaluation by MCTS wrapping. We present results for agents that learn Rubik's cube from scratch, with and without MCTS wrapping, with and without symmetries and show that both, MCTS wrapping and symmetries, increase computational costs, but lead at the same time to much better results. We can solve the 2x2x2 cube completely, and the 3x3x3 cube in the majority of the cases for scrambled cubes up to p = 15 (QTM). We cannot yet reliably solve 3x3x3 cubes with more than 15 scrambling twists. Although our computational costs are higher with MCTS wrapping and with symmetries than without, they are still considerably lower than in the approaches of McAleer et al. (2018, 2019) and Agostinelli et al. (2019) who provide the best Rubik's cube learning agents so far.

The generation of SAT instances is an important issue in computer science, and it is useful for researchers to verify the effectiveness of SAT solvers. Addressing this issue could inspire researchers to propose new search strategies. SAT problems exist in various real-world applications, some of which have more than one solution. However, although several algorithms for generating random SAT instances have been proposed, few can be used to generate hard instances that have multiple predefined solutions. In this paper, we propose the KHidden-M algorithm to generate SAT instances with multiple predefined solutions that could be hard to solve by the local search strategy when the number of predefined solutions is small enough and the Hamming distance between them is not less than half of the solution length. Specifically, first, we generate an SAT instance that is satisfied by all of the predefined solutions. Next, if the generated SAT instance does not satisfy the hardness condition, then a strategy will be conducted to adjust clauses through multiple iterations to improve the hardness of the whole instance. We propose three strategies to generate the SAT instance in the first part. The first strategy is called the random strategy, which randomly generates clauses that are satisfied by all of the predefined solutions. The other two strategies are called the estimating strategy and greedy strategy, and using them, we attempt to generate an instance that directly satisfies or is closer to the hardness condition for the local search strategy. We employ two SAT solvers (i.e., WalkSAT and Kissat) to investigate the hardness of the SAT instances generated by our algorithm in the experiments. The experimental results show the effectiveness of the random, estimating and greedy strategies. Compared to the state-of-the-art algorithm for generating SAT instances with predefined solutions, namely, M-hidden, our algorithm could be more effective in generating hard SAT instances.

The reaction center consists of atoms in the product whose local properties are not identical to the corresponding atoms in the reactants. Prior studies on reaction center identification are mainly on semi-templated retrosynthesis methods. Moreover, they are limited to single reaction center identification. However, many reaction centers are comprised of multiple bonds or atoms in reality. We refer to it as the multiple reaction center. This paper presents RCsearcher, a unified framework for single and multiple reaction center identification that combines the advantages of the graph neural network and deep reinforcement learning. The critical insight in this framework is that the single or multiple reaction center must be a node-induced subgraph of the molecular product graph. At each step, it considers choosing one node in the molecular product graph and adding it to the explored node-induced subgraph as an action. Comprehensive experiments demonstrate that RCsearcher consistently outperforms other baselines and can extrapolate the reaction center patterns that have not appeared in the training set. Ablation experiments verify the effectiveness of individual components, including the beam search and one-hop constraint of action space.

In this paper, we study the \underline{R}obust \underline{o}ptimization for \underline{se}quence \underline{Net}worked \underline{s}ubmodular maximization (RoseNets) problem. We interweave the robust optimization with the sequence networked submodular maximization. The elements are connected by a directed acyclic graph and the objective function is not submodular on the elements but on the edges in the graph. Under such networked submodular scenario, the impact of removing an element from a sequence depends both on its position in the sequence and in the network. This makes the existing robust algorithms inapplicable. In this paper, we take the first step to study the RoseNets problem. We design a robust greedy algorithm, which is robust against the removal of an arbitrary subset of the selected elements. The approximation ratio of the algorithm depends both on the number of the removed elements and the network topology. We further conduct experiments on real applications of recommendation and link prediction. The experimental results demonstrate the effectiveness of the proposed algorithm.

Conventional rule learning algorithms aim at finding a set of simple rules, where each rule covers as many examples as possible. In this paper, we argue that the rules found in this way may not be the optimal explanations for each of the examples they cover. Instead, we propose an efficient algorithm that aims at finding the best rule covering each training example in a greedy optimization consisting of one specialization and one generalization loop. These locally optimal rules are collected and then filtered for a final rule set, which is much larger than the sets learned by conventional rule learning algorithms. A new example is classified by selecting the best among the rules that cover this example. In our experiments on small to very large datasets, the approach's average classification accuracy is higher than that of state-of-the-art rule learning algorithms. Moreover, the algorithm is highly efficient and can inherently be processed in parallel without affecting the learned rule set and so the classification accuracy. We thus believe that it closes an important gap for large-scale classification rule induction.

Planning as heuristic search is one of the most successful approaches to classical planning but unfortunately, it does not extend trivially to Generalized Planning (GP). GP aims to compute algorithmic solutions that are valid for a set of classical planning instances from a given domain, even if these instances differ in the number of objects, the number of state variables, their domain size, or their initial and goal configuration. The generalization requirements of GP make it impractical to perform the state-space search that is usually implemented by heuristic planners. This paper adapts the planning as heuristic search paradigm to the generalization requirements of GP, and presents the first native heuristic search approach to GP. First, the paper introduces a new pointer-based solution space for GP that is independent of the number of classical planning instances in a GP problem and the size of those instances (i.e. the number of objects, state variables and their domain sizes). Second, the paper defines a set of evaluation and heuristic functions for guiding a combinatorial search in our new GP solution space. The computation of these evaluation and heuristic functions does not require grounding states or actions in advance. Therefore our GP as heuristic search approach can handle large sets of state variables with large numerical domains, e.g.~integers. Lastly, the paper defines an upgraded version of our novel algorithm for GP called Best-First Generalized Planning (BFGP), that implements a best-first search in our pointer-based solution space, and that is guided by our evaluation/heuristic functions for GP.