Showing 1–8 of 8 results for author: Morrill, J

New directions in the applications of rough path theory

Authors: Adeline Fermanian, Terry Lyons, James Morrill, Cristopher Salvi

Abstract: This article provides a concise overview of some of the recent advances in the application of rough path theory to machine learning. Controlled differential equations (CDEs) are discussed as the key mathematical model to describe the interaction of a stream with a physical control system. A collection of iterated integrals known as the signature naturally arises in the description of the response… ▽ More This article provides a concise overview of some of the recent advances in the application of rough path theory to machine learning. Controlled differential equations (CDEs) are discussed as the key mathematical model to describe the interaction of a stream with a physical control system. A collection of iterated integrals known as the signature naturally arises in the description of the response produced by such interactions. The signature comes equipped with a variety of powerful properties rendering it an ideal feature map for streamed data. We summarise recent advances in the symbiosis between deep learning and CDEs, studying the link with RNNs and culminating with the Neural CDE model. We concluded with a discussion on signature kernel methods. △ Less

Submitted 9 February, 2023; originally announced February 2023.

Neural Controlled Differential Equations for Online Prediction Tasks

Authors: James Morrill, Patrick Kidger, Lingyi Yang, Terry Lyons

Abstract: Neural controlled differential equations (Neural CDEs) are a continuous-time extension of recurrent neural networks (RNNs), achieving state-of-the-art (SOTA) performance at modelling functions of irregular time series. In order to interpret discrete data in continuous time, current implementations rely on non-causal interpolations of the data. This is fine when the whole time series is observed in… ▽ More Neural controlled differential equations (Neural CDEs) are a continuous-time extension of recurrent neural networks (RNNs), achieving state-of-the-art (SOTA) performance at modelling functions of irregular time series. In order to interpret discrete data in continuous time, current implementations rely on non-causal interpolations of the data. This is fine when the whole time series is observed in advance, but means that Neural CDEs are not suitable for use in \textit{online prediction tasks}, where predictions need to be made in real-time: a major use case for recurrent networks. Here, we show how this limitation may be rectified. First, we identify several theoretical conditions that interpolation schemes for Neural CDEs should satisfy, such as boundedness and uniqueness. Second, we use these to motivate the introduction of new schemes that address these conditions, offering in particular measurability (for online prediction), and smoothness (for speed). Third, we empirically benchmark our online Neural CDE model on three continuous monitoring tasks from the MIMIC-IV medical database: we demonstrate improved performance on all tasks against ODE benchmarks, and on two of the three tasks against SOTA non-ODE benchmarks. △ Less

Submitted 21 June, 2021; originally announced June 2021.

Neural Rough Differential Equations for Long Time Series

Authors: James Morrill, Cristopher Salvi, Patrick Kidger, James Foster, Terry Lyons

Abstract: Neural controlled differential equations (CDEs) are the continuous-time analogue of recurrent neural networks, as Neural ODEs are to residual networks, and offer a memory-efficient continuous-time way to model functions of potentially irregular time series. Existing methods for computing the forward pass of a Neural CDE involve embedding the incoming time series into path space, often via interpol… ▽ More Neural controlled differential equations (CDEs) are the continuous-time analogue of recurrent neural networks, as Neural ODEs are to residual networks, and offer a memory-efficient continuous-time way to model functions of potentially irregular time series. Existing methods for computing the forward pass of a Neural CDE involve embedding the incoming time series into path space, often via interpolation, and using evaluations of this path to drive the hidden state. Here, we use rough path theory to extend this formulation. Instead of directly embedding into path space, we instead represent the input signal over small time intervals through its \textit{log-signature}, which are statistics describing how the signal drives a CDE. This is the approach for solving \textit{rough differential equations} (RDEs), and correspondingly we describe our main contribution as the introduction of Neural RDEs. This extension has a purpose: by generalising the Neural CDE approach to a broader class of driving signals, we demonstrate particular advantages for tackling long time series. In this regime, we demonstrate efficacy on problems of length up to 17k observations and observe significant training speed-ups, improvements in model performance, and reduced memory requirements compared to existing approaches. △ Less

Submitted 21 June, 2021; v1 submitted 17 September, 2020; originally announced September 2020.

Comments: Published at ICML 2021

A Generalised Signature Method for Multivariate Time Series Feature Extraction

Authors: James Morrill, Adeline Fermanian, Patrick Kidger, Terry Lyons

Abstract: The 'signature method' refers to a collection of feature extraction techniques for multivariate time series, derived from the theory of controlled differential equations. There is a great deal of flexibility as to how this method can be applied. On the one hand, this flexibility allows the method to be tailored to specific problems, but on the other hand, can make precise application challenging.… ▽ More The 'signature method' refers to a collection of feature extraction techniques for multivariate time series, derived from the theory of controlled differential equations. There is a great deal of flexibility as to how this method can be applied. On the one hand, this flexibility allows the method to be tailored to specific problems, but on the other hand, can make precise application challenging. This paper makes two contributions. First, the variations on the signature method are unified into a general approach, the \emph{generalised signature method}, of which previous variations are special cases. A primary aim of this unifying framework is to make the signature method more accessible to any machine learning practitioner, whereas it is now mostly used by specialists. Second, and within this framework, we derive a canonical collection of choices that provide a domain-agnostic starting point. We derive these choices as a result of an extensive empirical study on 26 datasets and go on to show competitive performance against current benchmarks for multivariate time series classification. Finally, to ease practical application, we make our techniques available as part of the open-source [redacted] project. △ Less

Submitted 6 February, 2021; v1 submitted 1 June, 2020; originally announced June 2020.

Comments: 25 pages

Generalised Interpretable Shapelets for Irregular Time Series

Authors: Patrick Kidger, James Morrill, Terry Lyons

Abstract: The shapelet transform is a form of feature extraction for time series, in which a time series is described by its similarity to each of a collection of `shapelets'. However it has previously suffered from a number of limitations, such as being limited to regularly-spaced fully-observed time series, and having to choose between efficient training and interpretability. Here, we extend the method to… ▽ More The shapelet transform is a form of feature extraction for time series, in which a time series is described by its similarity to each of a collection of `shapelets'. However it has previously suffered from a number of limitations, such as being limited to regularly-spaced fully-observed time series, and having to choose between efficient training and interpretability. Here, we extend the method to continuous time, and in doing so handle the general case of irregularly-sampled partially-observed multivariate time series. Furthermore, we show that a simple regularisation penalty may be used to train efficiently without sacrificing interpretability. The continuous-time formulation additionally allows for learning the length of each shapelet (previously a discrete object) in a differentiable manner. Finally, we demonstrate that the measure of similarity between time series may be generalised to a learnt pseudometric. We validate our method by demonstrating its performance and interpretability on several datasets; for example we discover (purely from data) that the digits 5 and 6 may be distinguished by the chirality of their bottom loop, and that a kind of spectral gap exists in spoken audio classification. △ Less

Submitted 29 May, 2020; v1 submitted 28 May, 2020; originally announced May 2020.

Neural Controlled Differential Equations for Irregular Time Series

Authors: Patrick Kidger, James Morrill, James Foster, Terry Lyons

Abstract: Neural ordinary differential equations are an attractive option for modelling temporal dynamics. However, a fundamental issue is that the solution to an ordinary differential equation is determined by its initial condition, and there is no mechanism for adjusting the trajectory based on subsequent observations. Here, we demonstrate how this may be resolved through the well-understood mathematics o… ▽ More Neural ordinary differential equations are an attractive option for modelling temporal dynamics. However, a fundamental issue is that the solution to an ordinary differential equation is determined by its initial condition, and there is no mechanism for adjusting the trajectory based on subsequent observations. Here, we demonstrate how this may be resolved through the well-understood mathematics of \emph{controlled differential equations}. The resulting \emph{neural controlled differential equation} model is directly applicable to the general setting of partially-observed irregularly-sampled multivariate time series, and (unlike previous work on this problem) it may utilise memory-efficient adjoint-based backpropagation even across observations. We demonstrate that our model achieves state-of-the-art performance against similar (ODE or RNN based) models in empirical studies on a range of datasets. Finally we provide theoretical results demonstrating universal approximation, and that our model subsumes alternative ODE models. △ Less

Submitted 5 November, 2020; v1 submitted 18 May, 2020; originally announced May 2020.

Comments: Accepted at NeurIPS 2020 (Spotlight)

UV solar irradiance in observations and the NRLSSI and SATIRE-S models

Authors: K. L. Yeo, W. T. Ball, N. A. Krivova, S. K. Solanki, Y. C. Unruh, J. Morrill

Abstract: Total solar irradiance and UV spectral solar irradiance have been monitored since 1978 through a succession of space missions. This is accompanied by the development of models aimed at replicating solar irradiance by relating the variability to solar magnetic activity. The NRLSSI and SATIRE-S models provide the most comprehensive reconstructions of total and spectral solar irradiance over the peri… ▽ More Total solar irradiance and UV spectral solar irradiance have been monitored since 1978 through a succession of space missions. This is accompanied by the development of models aimed at replicating solar irradiance by relating the variability to solar magnetic activity. The NRLSSI and SATIRE-S models provide the most comprehensive reconstructions of total and spectral solar irradiance over the period of satellite observation currently available. There is persistent controversy between the various measurements and models in terms of the wavelength dependence of the variation over the solar cycle, with repercussions on our understanding of the influence of UV solar irradiance variability on the stratosphere. We review the measurement and modelling of UV solar irradiance variability over the period of satellite observation. The SATIRE-S reconstruction is consistent with spectral solar irradiance observations where they are reliable. It is also supported by an independent, empirical reconstruction of UV spectral solar irradiance based on UARS/SUSIM measurements from an earlier study. The weaker solar cycle variability produced by NRLSSI between 300 and 400 nm is not evident in any available record. We show that although the method employed to construct NRLSSI is principally sound, reconstructed solar cycle variability is detrimentally affected by the uncertainty in the SSI observations it draws upon in the derivation. Based on our findings, we recommend, when choosing between the two models, the use of SATIRE-S for climate studies. △ Less

Submitted 5 July, 2015; originally announced July 2015.

Journal ref: J. Geophys. Res. (Space Phys.) 120 (2015)

NUV/Blue spectral observations of sprites in the 320-460 nm region: ${\mathrm N_2}$ (2PG) Emissions

Authors: M. J. Heavner, J. S. Morrill, C. Siefring, D. D. Sentman, D. R. Moudry, E. M. Wescott, E. J. Bucsela

Abstract: A near-ultraviolet (NUV) spectrograph (320-460 nm) was flown on the EXL98 aircraft sprite observation campaign during July 1998. In this wavelength range video rate (60 fields/sec) spectrographic observations found the NUV/blue emissions to be predominantly N2 (2PG). The negligible level of N2+ (1NG) present in the spectrum is confirmed by observations of a co-aligned, narrowly filtered 427.8 nm… ▽ More A near-ultraviolet (NUV) spectrograph (320-460 nm) was flown on the EXL98 aircraft sprite observation campaign during July 1998. In this wavelength range video rate (60 fields/sec) spectrographic observations found the NUV/blue emissions to be predominantly N2 (2PG). The negligible level of N2+ (1NG) present in the spectrum is confirmed by observations of a co-aligned, narrowly filtered 427.8 nm imager and is in agreement with previous ground-based filtered photometer observations. The synthetic spectral fit to the observations indicates a characteristic energy of ~1.8 eV, in agreement with our other NUV observations. △ Less

Submitted 20 February, 2010; originally announced February 2010.

Comments: 7 pages, 2 figures, 1 table, JGR Space Physics "Effects of Thunderstorms and Lightning in the Upper Atmosphere" Special Section