Distribution Regression for Sequential Data
arXiv:2006.05805
Abstract
Distribution regression refers to the supervised learning problem where labels are only available for groups of inputs instead of individual inputs. In this paper, we develop a rigorous mathematical framework for distribution regression where inputs are complex data streams. Leveraging properties of the expected signature and a recent signature kernel trick for sequential data from stochastic analysis, we introduce two new learning techniques, one feature-based and the other kernel-based. Each is suited to a different data regime in terms of the number of data streams and the dimensionality of the individual streams. We provide theoretical results on the universality of both approaches and demonstrate empirically their robustness to irregularly sampled multivariate time-series, achieving state-of-the-art performance on both synthetic and real-world examples from thermodynamics, mathematical finance and agricultural science.
arXiv commentPublished at AISTATS 2021
References in corpus (17)
- Scikit-learn: Machine Learning in Python
- GPyTorch: Blackbox Matrix-Matrix Gaussian Process Inference with GPU Acceleration
- A kernel for time series based on global alignments
- Soft-DTW: a Differentiable Loss Function for Time-Series
- Persistence weighted Gaussian kernel for topological data analysis
- Learning from Distributions via Support Measure Machines
- Sparse arrays of signatures for online character recognition
- Rough paths, Signatures and the modelling of functions on streams
- Signatory: differentiable computations of the signature and logsignature transforms, on both CPU and GPU
- Deep Signature Transforms
- Signature moments to characterize laws of stochastic processes
- Using path signatures to predict a diagnosis of Alzheimer's disease
- Autoregressive Kernels For Time Series
- The iisignature library: efficient calculation of iterated-integral signatures and log signatures
- Multi-resolution Multi-task Gaussian Processes
- Variational Learning on Aggregate Outputs with Gaussian Processes
- Embedding and learning with signatures