A 1/2-Approximation for Budgeted $k$-Submodular Maximization

2025-07-17

AI Summary

This paper studies the budget-constrained $k$-submodular maximization problem, an NP-hard optimization task. We propose the 1-Guess Greedy algorithm: by guessing at most one critical element from the optimal solution in a single pass—and integrating greedy selection with threshold-based truncation—it achieves tight approximation ratios of $1/2$ for monotone and $1/3$ for non-monotone objectives, respectively; this is the first proof of a tight $1/2$-approximation bound under budget constraints. Technically, we introduce a novel continuous analysis framework based on the multilinear extension, which unifies treatment across diverse constraints while ensuring simplicity, parallelizability, and theoretical tightness. The algorithm runs in $ ilde{O}(n^2 k^2)$ time, substantially improving upon prior approaches.

Technology Category

Search and Optimization: Mixed Discrete/Continuous SearchConstraint Satisfaction and Optimization: Mixed Discrete/Continuous OptimizationReasoning under Uncertainty: Stochastic Optimization

Application Category

Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphsResponsible Web: Human-perceived consequences of algorithmic deployment on the webEconomics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

A $k$-submodular function naturally generalizes submodular functions by taking as input $k$ disjoint subsets, rather than a single subset. Unlike standard submodular maximization, which only requires selecting elements for the solution, $k$-submodular maximization adds the challenge of determining the subset to which each selected element belongs. Prior research has shown that the greedy algorithm is a 1/2-approximation for the monotone $k$-submodular maximization problem under cardinality or matroid constraints. However, whether a firm 1/2-approximation exists for the budgeted version (i.e., with a knapsack constraint) has remained open for several years. We resolve this question affirmatively by proving that the 1-Guess Greedy algorithm, which first guesses an appropriate element from an optimal solution before proceeding with the greedy algorithm, achieves a 1/2-approximation. This result is asymptotically tight as $((k+1)/(2k)+ε)$-approximation requires exponentially many value oracle queries even without constraints (Iwata et al., SODA 2016). We further show that 1-Guess Greedy is 1/3-approximation for the non-monotone problem. This algorithm is both simple and parallelizable, making it well-suited for practical applications. Using the thresholding technique from (Badanidiyuru and Vondrak, SODA 2014), it runs in nearly $ ilde O(n^2k^2)$ time. The proof idea is simple: we introduce a novel continuous transformation from an optimal solution to a greedy solution, using the multilinear extension to evaluate every fractional solution during the transformation. This continuous analysis approach yields two key extensions. First, it enables improved approximation ratios of various existing algorithms. Second, our method naturally extends to $k$-submodular maximization problems under broader constraints, offering a more flexible and unified analysis framework.

Problem

• Resolving 1/2-approximation for budgeted k-submodular maximization
• Extending analysis to non-monotone k-submodular functions
• Providing a unified framework for broader constraint settings

Innovation

• 1-Guess Greedy algorithm for 1/2-approximation
• Continuous transformation using multilinear extension
• Nearly linear time with thresholding technique