PTF Testing Lower Bounds for Non-Gaussian Component Analysis

2025-11-24

AI Summary

This work addresses the long-standing open problem of establishing nontrivial lower bounds for polynomial threshold function (PTF) tests in statistical tasks such as Non-Gaussian Component Analysis (NGCA), where such bounds were previously absent. Methodologically, it introduces a novel synthesis of pseudorandom generator theory and structured PTF analysis to characterize the behavioral limits of low-degree polynomials under random directions. The main contributions are: (1) deriving a near-optimal PTF lower bound for NGCA, thereby exposing an inherent information-computation tradeoff; (2) transcending the limitations of the standard low-degree polynomial method by enabling lower-bound analysis in more natural computational models; and (3) obtaining corollary lower bounds for several related statistical problems—including sparse PCA and Gaussian mixture classification—providing rigorous theoretical foundations for computational complexity tradeoffs in statistical learning.

Technology Category

• Machine Learning: Probabilistic Circuits and Graphical Models
• Reasoning under Uncertainty: Graphical Models
• Constraint Satisfaction and Optimization: Other Foundations of Constraint Satisfaction

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Security and Privacy: Large-scale security measurements
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

This work studies information-computation gaps for statistical problems. A common approach for providing evidence of such gaps is to show sample complexity lower bounds (that are stronger than the information-theoretic optimum) against natural models of computation. A popular such model in the literature is the family of low-degree polynomial tests. While these tests are defined in such a way that make them easy to analyze, the class of algorithms that they rule out is somewhat restricted. An important goal in this context has been to obtain lower bounds against the stronger and more natural class of low-degree Polynomial Threshold Function (PTF) tests, i.e., any test that can be expressed as comparing some low-degree polynomial of the data to a threshold. Proving lower bounds against PTF tests has turned out to be challenging. Indeed, we are not aware of any non-trivial PTF testing lower bounds in the literature. In this paper, we establish the first non-trivial PTF testing lower bounds for a range of statistical tasks. Specifically, we prove a near-optimal PTF testing lower bound for Non-Gaussian Component Analysis (NGCA). Our NGCA lower bound implies similar lower bounds for a number of other statistical problems. Our proof leverages a connection to recent work on pseudorandom generators for PTFs and recent techniques developed in that context. At the technical level, we develop several tools of independent interest, including novel structural results for analyzing the behavior of low-degree polynomials restricted to random directions.

Problem

• Establishes first PTF testing lower bounds for statistical tasks
• Proves near-optimal lower bounds for Non-Gaussian Component Analysis
• Develops novel tools for analyzing low-degree polynomial behavior

Innovation

• Establishes first PTF testing lower bounds
• Proves near-optimal bounds for Non-Gaussian Component Analysis
• Leverages pseudorandom generators and novel structural analysis