Semi-parametric Functional Classification via Path Signatures Logistic Regression with Adaptive Order Selection
arXiv:2507.06637v2 Announce Type: replace-cross Abstract: We propose Path Signatures Logistic Regression (PSLR), a semi-parametric framework for classifying vector-valued functional data with scalar covariates. Classical functional logistic regression models rely on linear assumptions and fixed basis expansions, which limit flexibility and degrade performance under irregular sampling. PSLR leverages the well-established properties of path signatures - basis-free representation, cross-channel dependency capture, and robustness to sampling irregularity - as an enabling tool. The key novelty, how
The continuous evolution of AI research seeks more robust and flexible methods for functional data analysis, addressing limitations of traditional statistical models.
This development offers a more nuanced and powerful approach to classifying complex functional data, which can enhance AI applications in diverse fields by improving predictive accuracy and handling data irregularities.
The introduction of Path Signatures Logistic Regression (PSLR) provides an alternative to existing functional classification models, potentially leading to more reliable AI systems that can manage irregular data more effectively.
- · Machine Learning Researchers
- · Data Scientists
- · AI-driven industries with complex time-series data
- · Developers relying solely on linear functional models
- · Systems unable to integrate new statistical methods
Improved performance in AI systems dealing with complex, irregular functional data.
Broader adoption of signature-based methods in machine learning for time-series and functional data.
Enhanced AI capabilities leading to new applications in fields like finance, healthcare, and engineering where functional data is prevalent.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG