A supervised feature selection method selects an appropriate but concise set of features to differentiate classes, which is highly expensive for large-scale datasets. Therefore, feature selection should aim at both minimizing the number of selected features and maximizing the accuracy of classification, or any other task. However, this crucial task is computationally highly demanding on many real-world datasets and requires a very efficient algorithm to reach a set of optimal features with a limited number of fitness evaluations. For this purpose, we have proposed the binary multi-objective coordinate search (MOCS) algorithm to solve large-scale feature selection problems. To the best of our knowledge, the proposed algorithm in this paper is the first multi-objective coordinate search algorithm. In this method, we generate new individuals by flipping a variable of the candidate solutions on the Pareto front. This enables us to investigate the effectiveness of each feature in the corresponding subset. In fact, this strategy can play the role of crossover and mutation operators to generate distinct subsets of features. The reported results indicate the significant superiority of our method over NSGA-II, on five real-world large-scale datasets, particularly when the computing budget is limited. Moreover, this simple hyper-parameter-free algorithm can solve feature selection much faster and more efficiently than NSGA-II.

First of all, there is a lack of systematic methods that improve the performance of algorithms on unseen instances by gathering the experience from the instances solved in the past. Therefore, all the information obtained during the past runs of an algorithm is neglected when a new instance is encountered. A good example is the branchand-bound-and-remember method [34, 40], where the algorithm remembers the information derived during an instance solving, but the information is forgotten as soon as the instance is solved. Second, the development of efficient heuristic rules requires a substantial amount of time devoted to its design and testing. This process is tedious and requires a skilled human professional to fine-tune the heuristic's parameters. A typical example of this feature is genetic algorithms having many parameters for selection, cross-over, mutation, and other operators. The apparent response to the above challenges is utilizing the existing data. However, the main obstacle to the successful application of machine learning to enhance algorithms for combinatorial problems remains. It can be formulated as the following fundamental question--is it possible to extract any useful information from the solved instances and use it efficiently to accelerate solving of an unseen instance?

Minimax optimization problems have attracted a lot of attention over the past few years, with applications ranging from economics to machine learning. While advanced optimization methods exist for such problems, characterizing their dynamics in stochastic scenarios remains notably challenging. In this paper, we pioneer the use of stochastic differential equations (SDEs) to analyze and compare Minimax optimizers. Our SDE models for Stochastic Gradient Descent-Ascent, Stochastic Extragradient, and Stochastic Hamiltonian Gradient Descent are provable approximations of their algorithmic counterparts, clearly showcasing the interplay between hyperparameters, implicit regularization, and implicit curvature-induced noise. This perspective also allows for a unified and simplified analysis strategy based on the principles of It\^o calculus. Finally, our approach facilitates the derivation of convergence conditions and closed-form solutions for the dynamics in simplified settings, unveiling further insights into the behavior of different optimizers.

Randomly initialized dense networks contain subnetworks that achieve high accuracy without weight learning -- strong lottery tickets (SLTs). Recently, Gadhikar et al. (2023) demonstrated theoretically and experimentally that SLTs can also be found within a randomly pruned source network, thus reducing the SLT search space. However, this limits the search to SLTs that are even sparser than the source, leading to worse accuracy due to unintentionally high sparsity. This paper proposes a method that reduces the SLT search space by an arbitrary ratio that is independent of the desired SLT sparsity. A random subset of the initial weights is excluded from the search space by freezing it -- i.e., by either permanently pruning them or locking them as a fixed part of the SLT. Indeed, the SLT existence in such a reduced search space is theoretically guaranteed by our subset-sum approximation with randomly frozen variables. In addition to reducing search space, the random freezing pattern can also be exploited to reduce model size in inference. Furthermore, experimental results show that the proposed method finds SLTs with better accuracy and model size trade-off than the SLTs obtained from dense or randomly pruned source networks. In particular, the SLT found in a frozen graph neural network achieves higher accuracy than its weight trained counterpart while reducing model size by $40.3\times$.

We study a noise model for linear stochastic bandits for which the subgaussian noise parameter vanishes linearly as we select actions on the unit sphere closer and closer to the unknown vector. We introduce an algorithm for this problem that exhibits a minimax regret scaling as $\log^3(T)$ in the time horizon $T$, in stark contrast the square root scaling of this regret for typical bandit algorithms. Our strategy, based on weighted least-squares estimation, achieves the eigenvalue relation $\lambda_{\min} ( V_t ) = \Omega (\sqrt{\lambda_{\max}(V_t ) })$ for the design matrix $V_t$ at each time step $t$ through geometrical arguments that are independent of the noise model and might be of independent interest. This allows us to tightly control the expected regret in each time step to be of the order $O(\frac1{t})$, leading to the logarithmic scaling of the cumulative regret.

Branched optimal transport (BOT) is a generalization of optimal transport in which transportation costs along an edge are subadditive. This subadditivity models an increase in transport efficiency when shipping mass along the same route, favoring branched transportation networks.

Graph neural architecture search (GraphNAS) has recently aroused considerable attention in both academia and industry. However, two key challenges seriously hinder the further research of GraphNAS. First, since there is no consensus for the experimental setting, the empirical results in different research papers are often not comparable and even not reproducible, leading to unfair comparisons. Secondly, GraphNAS often needs extensive computations, which makes it highly inefficient and inaccessible to researchers without access to large-scale computation. To solve these challenges, we propose NAS-Bench-Graph, a tailored benchmark that supports unified, reproducible, and efficient evaluations for GraphNAS.

Epoch-Greedy has the following properties: No knowledge of a time horizon T is necessary. The regret incurred by Epoch-Greedy is controlled by a sample complexity bound for a hypothesis class. Here S is the complexity term in a sample complexity bound for standard supervised learning.

This paper uses information-theoretic techniques to determine minimax rates for estimating nonparametric sparse additive regression models under high-dimensional scaling. The first term reflects the difficulty of performing \emph{subset selection} and is independent of the Hilbert space \Hilb; the second term \LowerRateSq is an \emph{\s-dimensional estimation} term, depending only on the low dimension \s but not the ambient dimension \pdim, that captures the difficulty of estimating a sum of \s univariate functions in the Hilbert space \Hilb . The minimax rates are compared with rates achieved by an \ell_1 -penalty based approach, it can be shown that a certain \ell_1 -based approach achieves the minimax optimal rate.

Generalized Linear Models (GLMs) and Single Index Models (SIMs) provide powerful generalizations of linear regression, where the target variable is assumed to be a (possibly unknown) 1-dimensional function of a linear predictor. In general, these problems entail non-convex estimation procedures, and, in practice, iterative local search heuristics are often used. Kalai and Sastry (2009) provided the first provably efficient method, the \emph{Isotron} algorithm, for learning SIMs and GLMs, under the assumption that the data is in fact generated under a GLM and under certain monotonicity and Lipschitz (bounded slope) constraints. However, to obtain provable performance, the method requires a fresh sample every iteration. In this paper, we provide algorithms for learning GLMs and SIMs, which are both computationally and statistically efficient.