---
title: Greedy Online Change Point Detection
url: https://www.emergentmind.com/topics/greedy-online-change-point-detection-gocpd
type: topic
---

# Greedy Online Change Point Detection

Greedy Online Change Point Detection (GOCPD) refers to a class of online, data-adaptive algorithms that detect abrupt distributional changes (“change points”) in streaming or sequentially observed data through localized, computationally efficient, typically likelihood-based criteria and grid-scanning or segmentation strategies. These procedures are characterized by the greedy maximization of evidence for a change—often by exploiting unimodality, strong log-likelihood contrasts, or CUSUM-like substructure—while maintaining low computational and storage costs suitable for high-throughput, high-dimensional, or ill-conditioned environments.

## 1. Formal Problem Setting and GOCPD Objective

The canonical GOCPD problem considers an infinite or streaming sequence of observations (possibly multivariate or structured), with an unknown change point $\tau$ at which the underlying data-generating process transitions from a pre-change regime (with distribution $P_1$ or parameter $\theta_1$) to a post-change regime ($P_2$ or $\theta_2$). Denoting the observed data as $\mathcal{D} = \{Y_t\}_{t=1}^T$, the task is to design an online stopping rule $\widehat\tau$ such that, with high confidence and low delay,
- $\widehat\tau \approx \tau$ whenever a change occurs, and
- false alarms—detections before the true $\tau$ or in the absence of change—are controlled at a nominal level.

In the “greedy” GOCPD paradigm, at each time $t$, a candidate split point $c_t$ is identified by maximizing a criterion favoring segmentation into two independent regimes. For example, one widely used objective is
$$
c_t = \arg\max_{\tau} \left\{ \log\bar{p}(\mathcal{D}_{\tilde c(t):\tau-1} | \hat\theta_1^{(\tau)}) + \log\bar{p}(\mathcal{D}_{\tau:t} | \hat\theta_2^{(\tau)}) \right\}
$$
where $\bar{p}$ denotes the average log-likelihood under maximum likelihood parameters fitted to respective segments, and $\tilde c(t)$ is the last detected change point [2308.07012]. This framework naturally subsumes classical CUSUM, likelihood-ratio, and residual-based tests, but is operationalized online through computationally efficient greedy search or grid-based scanning.

## 2. Algorithmic Structure and Computational Efficiency

GOCPD methods achieve efficiency and online deployment primarily through (a) greedy/local scans over a dynamically maintained set of candidate change points, (b) judicious use of summary statistics and recurrence relations, and (c) objective functions admitting (piecewise) unimodality, enabling accelerated search.

### Greedy Search by Unimodality

For univariate or multivariate time series with a single change, the GOCPD objective is typically unimodal in the candidate index $\tau$—a formal property established in [2308.07012, Proposition 1]. This property allows the change point search to be performed via ternary search, reducing per-step computational cost from $O(t)$ to $O(\log t)$.

### Dynamic Geometric Grid

For large-scale or high-frequency settings, [2504.09573] proposes maintaining and updating a dynamically selected geometric grid $G^{(t)}$ of candidate offsets, with $|G^{(t)}|=O(\log t)$, guaranteeing that for any true jump $d\leq t/2$, there exists a grid point close to $d$. This enables grid-scanning CUSUM or likelihood-type tests to be performed in $O(\log t)$ time and space:
- At each $t$, all sufficient summaries or statistics for $g\in G^{(t)}$ are incrementally updated.
- For each $g$, a test statistic $T_g^{(t)}$ is computed; detection is triggered if any $T_g^{(t)}$ crosses a threshold.

### Memory and Storage

For high-dimensional data, summaries (such as partial sums or outer products) are only maintained for $O(\log t)$ relevant intervals, and many GOCPD architectures exploit tail-length or excitationset sparsity to further reduce redundancy [2003.03668, 2504.09573].

## 3. Test Statistics, Aggregation, and Robustification

GOCPD platforms support a wide variety of change detection statistics, including but not limited to:
- Likelihood-ratio and average log-likelihood–based scores [2308.07012]
- Multiscale, coordinate-wise CUSUM or CUSUM-like statistics and their cross-coordinate aggregations (for high-dimensions or signals of unknown sparsity) [2003.03668]
- Covariance, operator-norm, or residual-based metrics for detecting parameter or structural breaks [2504.09573, 2406.10349]

Robustification is achieved via post-split statistical confirmation, such as Mahalanobis screening of left- and right-segment residuals to guard against outlier-induced false positives. Empirically, these outlier guards reduce the false discovery rate (FDR) significantly in both synthetic and real-world benchmarks [2308.07012].

## 4. Theoretical Guarantees and Performance Metrics

GOCPD schemes are designed with explicit statistical and computational guarantees:
- **False Alarm Control:** The probability of detecting a change before a true change (or in its absence), $\mathrm{FA}(\widehat\tau)$, is controlled at level $\leq \delta$ for user-specified $\delta$ ([2504.09573, 2003.03668]).
- **Detection Delay:** For sufficiently large change magnitude $\phi=|\mu_2-\mu_1|$ (or appropriate high-dimensional analogs), the expected delay $\widehat{\tau}-\tau$ is provably $O(\log(\tau/\delta)/\phi^2)$, which matches information-theoretic minimax lower bounds up to logarithmic factors [2504.09573].
- **Computational Cost:** Update and storage cost per time step is $O(\log t)$ for scalar/low-dimensional tests, and $O(p\log p\,\log t)$ or better for multivariate mean and covariance detection [2504.09573, 2003.03668].
- **Empirical Performance:** On real and synthetic datasets, GOCPD yields true positive rates (TPR) in the range $0.86$–$1.00$ and positive predictive values (PPV) $0.67$–$0.83$, outperforming or matching established baselines in FDR and runtime [2308.07012].

## 5. Extensions: Greedy Excitation in System Identification

In adaptive and system identification settings with time-varying parameters, GOCPD is paired with *greedy excitation-set selection* for robust recursive least squares (RLS) updates. The key innovation is to maintain an online "greedy excitation set" $E_g(k)$: newly acquired regressors are admitted if and only if they do not worsen the Hessian condition number [2406.10349]. The parameter update at each step then uses a two-tier weighting—retaining informative past data and exponentially forgetting the rest—leading to improved tracking and bias-variance control.

An embedded GOCPD change point detector, based on EWMA-filtered model residuals and a likelihood-ratio test for jump-induced miss distributions, triggers a memory reset, discarding obsolete historical data and reinitializing the model post-jump for rapid reacquisition of new dynamics. This memory-resetting is provably optimal under likelihood tests and preserves adaptivity in ill-conditioned regimes.

## 6. Multiscale and High-Dimensional Frameworks

GOCPD architectures are designed to perform efficient detection in both low- and high-dimensional regimes. For high-dimensional Gaussian streams, multiscale likelihood-ratio and CUSUM-like statistics are computed over dyadic grids for each coordinate, yielding aggregation strategies (e.g., $\ell_2$-hard-thresholded sums) that adapt to varying sparsity levels and unknown signal strengths [2003.03668, 2504.09573]. All core formulas—per-coordinate scan statistics, aggregation, and stopping rules—are updated greedily with minimal memory and are implemented in software such as the R package 'ocd'.

The general methodology allows any offline scan statistic (CUSUM, LR, operator-norm, etc.) to be dynamically embedded into a streaming GOCPD architecture, transforming classical global scans into locally greedy, real-time detectors.

## 7. Practical Implementation and Applications

Published implementations emphasize the following features:
- Maintenance of summary statistics and candidate grids with $O(\log t)$ amortized time and space [2504.09573]
- Flexible modeling choices (Gaussian, GP, or regression models) according to domain requirements [2308.07012, 2406.10349]
- Outlier-robust postprocessing and statistical threshold calibration via theoretical/empirical procedures
- Empirically validated performance in applications ranging from high-frequency market data, EEG seizure detection, and activity monitoring to online adaptive regression in time-varying systems [2308.07012, 2504.09573, 2003.03668, 2406.10349]

A tabular synopsis of GOCPD methodologies drawn from key references follows:

| Methodology Source   | Core Idea                         | Per-step Complexity      |
|---------------------|-----------------------------------|-------------------------|
| [2308.07012]        | Unimodal loglikelihood maximization + ternary search | $O(\log t)$           |
| [2504.09573]        | Dynamic geometric scan grid, any offline statistic  | $O(\log t)$–$O(p^3\log t)$ |
| [2003.03668]        | Multiscale, coordinatewise greedy CUSUM + aggregation | $O(p^2 \log p)$       |
| [2406.10349]        | Greedy excitation-set RLS + LR change detection  | $O(p^2)$              |

The empirical and theoretical analyses collectively establish GOCPD as an efficient, versatile, and statistically rigorous framework for rapid, robust change-point detection under streaming constraints and challenging data regimes.

Source: https://www.emergentmind.com/topics/greedy-online-change-point-detection-gocpd