Bayesian Online Changepoint Detection
read the original abstract
Changepoints are abrupt variations in the generative parameters of a data sequence. Online detection of changepoints is useful in modelling and prediction of time series in application areas such as finance, biometrics, and robotics. While frequentist methods have yielded online filtering and prediction techniques, most Bayesian papers have focused on the retrospective segmentation problem. Here we examine the case where the model parameters before and after the changepoint are independent and we derive an online algorithm for exact inference of the most recent changepoint. We compute the probability distribution of the length of the current ``run,'' or time since the last changepoint, using a simple message-passing algorithm. Our implementation is highly modular so that the algorithm may be applied to a variety of types of data. We illustrate this modularity by demonstrating the algorithm on three different real-world data sets.
This paper has not been read by Pith yet.
Forward citations
Cited by 31 Pith papers
-
The Statistical Cost of Adaptation in Multi-Source Transfer Learning
Multi-source transfer learning incurs an intrinsic adaptation cost that can exceed one, with phase transitions separating regimes where bias-agnostic estimators match oracle performance from those where they cannot.
-
Gradient-Free Warm-Start Library Recovery: an Amortized-Regret Separation
A gradient-free warm-start library learner attains an amortized-regret separation on recurring-regime streams with recognition cost independent of dimension D.
-
A Switching Beamformer for Highly Non-Stationary Environments
Introduces the Universal Switching Beamformer that maintains an exponentially large set of covariance histories via a linear transition diagram and re-weights them by cumulative output power, with a proven regret boun...
-
Does Capability Transfer to Subjective Behavior -- and Would Our Instruments Tell Us? A Self-Evolving, Trust-by-Construction Evaluation Paradigm
Self-evolving rubric with anti-gaming fitness reveals that objective capability scaling fails to transfer to subjective LLM behaviors, with advice-restraint as the universal lowest dimension that can regress.
-
Generalized Stochastic Approximation of the Log-Likelihood Ratio for Robust Sequential Change-Point Detection
A new approximation method for the log-likelihood ratio allows robust sequential change-point detection in non-Gaussian processes using moments up to order 3s.
-
Geometric Observables for Financial Regime Detection
Geometric observables from spectral embeddings of stock returns detect financial regime shifts with competitive out-of-sample performance and fewer false alarms than supervised baselines.
-
BAPR: Bayesian amnesic piecewise-robust reinforcement learning for non-stationary continuous control
BAPR combines Bayesian change detection with robust RL, proves the core operator is a contraction via Lean 4, and adapts conservatism after detected regime shifts in continuous control.
-
Mixed neural posterior estimation for simulators with discrete and continuous parameters
Extends NPE to mixed discrete-continuous parameter spaces via a factorized inference network combining an autoregressive classifier and generative model, trained jointly to yield accurate calibrated posteriors.
-
CHASM: Online Changepoint Detection in Temporal and Cross-Variable Dependence
CHASM detects changes in temporal and cross-variable dependence in multivariate time series by monitoring the truncated eigenvalue sequence of a recursively estimated DMD operator, using optimal assignment and augment...
-
Early Detection of Latent Microstructure Regimes in Limit Order Books
A three-regime causal model with a latent build-up phase enables a MAX-aggregation trigger detector to deliver positive expected lead time before observable stress in limit order books.
-
On the Necessity of a Liquid Substrate for Mesh Intelligence
Continuous-time liquid networks are necessary for optimal estimation under irregular exogenous observations and fixed weights in mesh intelligence.
-
Return-to-Baseline Testing via Empirically Calibrated e-processes
A distribution-free sequential test for return-to-baseline detection that aggregates a universal-inference discrepancy into a super-martingale and empirically calibrates it into an e-process with finite-sample bounds.
-
SKIP: Sparse Keyframe Interpolation Paradigm for Efficient Embodied World Models
SKIP achieves 4.16x faster dense video rollouts for robot world models by synthesizing only multimodal-identified keyframes and interpolating the rest, preserving policy training effectiveness with minimal success rate drops.
-
SWORD: Spectral Wasserstein Online Regime Detection in Dynamic Networks
SWORD detects change points in dynamic graphs by averaging Chebyshev moments of the normalized Laplacian over two time windows and using L1 distance, improving mean F1 from 0.27 to 0.79 over prior spectral methods on ...
-
CALIBURN: Operationally Calibrated Streaming Intrusion Detection with Regime-Dependent Conformal Risk Control
CALIBURN integrates Bayesian change-point detection, isotonic calibration, cost-sensitive thresholding, conformal risk control, and burn-rate alerting into a single streaming substrate, showing that calibration and CR...
-
DeRegiME: Deep Regime Mixtures for Probabilistic Forecasting under Distribution Shift
DeRegiME uses a sparse variational GP with nonstationary regime-mixing kernel to decompose forecasts into mean, residual regimes, and noise for improved probabilistic forecasting under distribution shift.
-
Geometric Observables for Financial Regime Detection
Berry Phase Rate and other geometric observables from learned spectral embeddings of equity-index returns detect financial regime shifts with competitive out-of-sample performance and lower false-alarm rates than supe...
-
Online Bayesian Calibration under Gradual and Abrupt System Changes
BRPC is an online Bayesian calibration framework that decouples parameter tracking from discrepancy modeling for gradual nonstationarity and adds restart mechanisms to handle abrupt regime shifts.
-
Sequential Bayesian Monitoring for Recoverable and Drifting Processes
Bayesian procedures are derived to compute the posterior probability that a recoverable process is currently in control or that a drifting latent parameter lies in an acceptable region.
-
Pi-Change: A Prior-Informed Multiple Change Point Detection Algorithm
Pi-Change extends the PELT framework for multiple change point detection by incorporating prior information on locations through a time-varying penalty that preserves dynamic programming efficiency.
-
Cost-sensitive retraining via posterior learning debt
Posterior learning debt enables cost-sensitive retraining decisions that outperform calendar-based and CUSUM methods in synthetic Bayesian simulations.
-
A Path Signature Framework for Detecting Creative Fatigue in Digital Advertising
The paper introduces a path signature-based method for detecting creative fatigue in digital ads using geometric change detection on performance paths.
-
Continuous Hidden Markov Models for Equity Returns: Heavy-Tail Emission Families and Regime-Conditional Value-at-Risk
Heavy-tailed continuous HMMs recover volatility clustering and produce regime-conditional VaR that passes joint conditional coverage tests on US equity data.
-
Exploring Statistical Change Point Detection Techniques for Performance Anomaly Detection at Mozilla
Ensemble voting strategies for change point detection improve F1-score by 11% over Mozilla's T-test method on a new ground-truth dataset of 174 performance time series annotated by practitioners.
-
Anomaly Detection and Root Cause Analysis for Microservice Systems
Thesis proposes BARO for metrics, EventADL for events, TORAI for multimodal RCA without call graphs, and RCAEval benchmark with systematic evaluation of causal methods.
-
CHRONOS: Temporally-Aware Multi-Agent Coordination for Evolving Data Marketplaces
CHRONOS is a three-layer system for evolving data marketplaces that applies neural-ODE temporal decay, changepoint-aware Shapley valuation, and EXP3-IX private coordination to achieve 0.937 recall, 2.74 qps, 161 ms la...
-
Generalized Stochastic Approximation of the Log-Likelihood Ratio for Robust Sequential Change-Point Detection
A unified approximation framework for the log-likelihood ratio on polynomial/logarithmic/fractional-power bases using moments up to order 3s adapts CUSUM/GRSh/SRP procedures to non-Gaussian change-point detection and ...
-
Progressive Autonomy as Preference Learning: A Formalization of Trust Calibration for Agentic Tool Use
Trust calibration in agentic tool use is cast as preferential Bayesian optimization over a latent human risk-tolerance function observed through binary approve/deny feedback with a probit likelihood.
-
Safe Continual Reinforcement Learning under Nonstationarity via Adaptive Safety Constraints
LILAC+ combines context-based, adaptation-speed, and budget-to-state safety constraints to reduce violations in continual RL under nonstationary conditions, demonstrated in simulated driving tasks.
-
Data-driven moving-window Bayesian inference for transient CO2-temperature network models of buildings
A moving-window Bayesian inference procedure jointly estimates thermal parameters, airflow, occupancy trajectories, and sensor noise in a coupled CO2-temperature RC network model for buildings, achieving accurate traj...
-
Dynamic time series clustering via volatility change-points
A Bayesian method clusters time series by similarity in the timing of their most recent volatility change-points via a metric on posterior distributions, demonstrated on S&P 500 returns.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.