paper-with-me

Papers

pySigLib -- Fast Signature-Based Computations on CPU and GPU

2025-09-12 · Daniil Shmelev, Cristopher Salvi arxiv

Signature-based methods have recently gained significant traction in machine learning for sequential data. In particular, signature kernels have emerged as powerful discriminators and training losses for generative models on time-series, notably in quantitative finance. However, existing implementations do not scale to the dataset sizes and sequence lengths encountered in practice. We present pySigLib, a high-performance Python library offering optimised implementations of signatures and signature kernels on CPU and GPU, fully compatible with PyTorch's automatic differentiation. Beyond an efficient software stack for large-scale signature-based computation, we introduce a novel differentiation scheme for signature kernels that delivers accurate gradients at a fraction of the runtime of existing libraries.

📄 PDF Abstract BibTeX arXiv:2509.10613

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Random Controlled Differential Equations

2025-12-29 · Francesco Piatti, Thomas Cass, William F. Turner arxiv

We introduce a training-efficient framework for time-series learning that combines random features with controlled differential equations (CDEs). In this approach, large randomly parameterized CDEs act as continuous-time…

Fast Deep Stereo with 2D Convolutional Processing of Cost Signatures

2019-03-08 · Kyle Yee, Ayan Chakrabarti

Modern neural network-based algorithms are able to produce highly accurate depth estimates from stereo image pairs, nearly matching the reliability of measurements from more expensive depth sensors. However, this accurac…

Autonomous VehiclesGPU

Signatory: differentiable computations of the signature and logsignature transforms, on both CPU and GPU

2020-01-03 · ICLR 2021 1 · Patrick Kidger, Terry Lyons

Signatory is a library for calculating and performing functionality related to the signature and logsignature transforms. The focus is on machine learning, and as such includes features such as CPU parallelism, GPU suppo…

CPUGPU

MERCURY: Accelerating DNN Training By Exploiting Input Similarity

2021-10-28 · Vahid Janfaza, Kevin Weston, Moein Razavi, Shantanu Mandal 외

Deep Neural Networks (DNN) are computationally intensive to train. It consists of a large number of multidimensional dot products between many weights and input vectors. However, there can be significant similarity among…

Quantization

Emergent Riemannian geometry over learning discrete computations on continuous manifolds

2025-11-28 · Julian Brandon, Angus Chadwick, Arthur Pellegrino arxiv

Many tasks require mapping continuous input data (e.g. images) to discrete task outputs (e.g. class labels). Yet, how neural networks learn to perform such discrete computations on continuous data manifolds remains poorl…