Domain Showcase
Overview
Section titled “Overview”These applications demonstrate CVX’s generality across domains beyond the primary research areas. They are exploratory — proof-of-concept demonstrations using synthetic or benchmark data, not production-validated pipelines.
Available Showcases
Section titled “Available Showcases”Quantitative Finance
Section titled “Quantitative Finance”Market regime detection using temporal trajectory analysis. Embeddings of price/volume features track regime transitions (bull/bear/sideways). CVX detects change points and measures drift between regimes.
CVX features: detect_changepoints(), drift(), velocity(), region_trajectory()
Anomaly Detection (NAB)
Section titled “Anomaly Detection (NAB)”Evaluation on the Numenta Anomaly Benchmark. CVX’s change point detection (PELT/BOCPD) applied to time-series embedding trajectories for unsupervised anomaly detection.
CVX features: detect_changepoints(), bocpd_observe(), event_features()
Quality-Diversity (MAP-Elites)
Section titled “Quality-Diversity (MAP-Elites)”Using CVX as a behavioral archive for MAP-Elites evolutionary optimization. Embedding-based behavioral descriptors enable continuous behavioral spaces with HNSW-based niche assignment.
CVX features: regions(), region_assignments(), search(), path_signature()
MLOps Drift Detection
Section titled “MLOps Drift Detection”Monitoring embedding model drift in production ML systems. CVX tracks how model outputs evolve across retraining cycles, detecting concept drift and measuring distribution shift.
CVX features: cohort_drift(), wasserstein_drift(), fisher_rao_distance()
Molecular Dynamics & Drug Discovery
Section titled “Molecular Dynamics & Drug Discovery”Trajectory analysis for molecular conformational dynamics and binding site evolution. Conceptual application of CVX’s temporal primitives to molecular embedding spaces.
CVX features: trajectory(), velocity(), frechet_distance(), topological_features()
Fraud Detection & Insider Threat
Section titled “Fraud Detection & Insider Threat”Behavioral trajectory analysis for detecting anomalous patterns in user activity embeddings. Change point detection identifies behavioral regime changes.
CVX features: detect_changepoints(), discover_discords(), event_features()
Notebooks
Section titled “Notebooks”| Notebook | Domain | Data |
|---|---|---|
| T_Finance_Regimes | Market regimes | Synthetic |
| T_NAB_Anomaly | Anomaly detection | NAB benchmark |
| T_MAP_Elites | Quality-diversity | Synthetic |
| T_MLOps_Drift | Model monitoring | Synthetic |
| T_Fraud_Detection | Fraud detection | Synthetic |
| T_Insider_Threat | Insider threat | Synthetic |