arXiv:2505. 04608v5 Announce Type: replace-cross Abstract: Responsibly deploying artificial intelligence (AI) / machine learning (ML) systems in high-stakes settings arguably requires not only proof of system reliability, but also continual, post-deployment monitoring to quickly detect and address any unsafe behavior.
arXiv:2606. 09430v1 Announce Type: cross Abstract: Online task-free continual learning (TFCL) requires intelligent agents to sequentially accumulate knowledge from an unbounded, non-stationary data stream under strict single-pass constraints and without any explicit task identifiers.
The paper introduces a test‑time adaptation (TTA) method for AI text detection that addresses continual distribution shifts such as adversarial humanization, new large language models, and temporal drift in human writing. By exploiting inference‑time homogeneity among unlabeled samples, the TTA approach achieves robust performance, outperforming state‑of‑the‑art supervised detectors—e.g., improving detection of adversarial AI‑generated text from 27.0% to 93.3%. The study demonstrates that TTA can maintain high detection accuracy in real‑world, evolving deployment scenarios.
arXiv:2609.24278v1 Announce Type: new Abstract: Change point detection (CPD) identifies abrupt and significant changes in sequential data, with applications in human activity recognition, financial m...
arXiv:2606. 07789v1 Announce Type: new Abstract: Data stream mining is fundamentally challenged by concept drift, where distributional changes can degrade model performance.
The paper introduces the Stream Cruise Control Method (SCCM), a framework for detecting and adapting to concept drift in online regression. SCCM performs early-response drift detection, quantifies drift magnitude, applies KPI-window-based thresholding to reduce false alarms, dynamically tunes hyperparameters, and recalibrates models, all within an in-memory design for real-time operation. Evaluations on synthetic and real-world datasets demonstrate that SCCM improves predictive performance compared to eight baseline detector–adaptation methods.
The paper introduces an adaptive triggering mechanism for bias correction in large language model (LLM) reasoning. By framing bias intervention as an online change‑point detection problem, the authors update a CUSUM statistic at each step using either a white‑box next‑token probability signal or a black‑box LLM judge signal, and inject corrective prompts only when the accumulated evidence exceeds a calibrated threshold. Experiments on gpt‑4o‑mini and six open‑weight models show that adaptive black‑box triggering restores most of the accuracy lost by fixed‑interval interventions while reducing the number of corrections, whereas the white‑box signal improves ambiguous‑item accuracy but can hurt disambiguated‑item accuracy due to difficulty distinguishing stereotype reliance from correct evidence.
arXiv:2607. 18899v1 Announce Type: new Abstract: Forecasting under real-world conditions is inherently non-stationary, as the conditional distribution of future observations evolves over time.
MORA is a drift‑robust time‑series anomaly detection framework that addresses the ambiguity between distribution drift and true anomalies by defining a temporal change disambiguation problem. It reconstructs a local target from paired short‑ and long‑term views, using the reconstruction gap as a measure of contextual support to adjust the primary anomaly score. The method requires no drift annotations or online adaptation and demonstrates strong robustness to non‑stationarity while maintaining sensitivity to genuine anomalies on four TSAD benchmarks.
arXiv:2606. 20216v1 Announce Type: cross Abstract: Machine learning algorithms deployed for evolving streaming environments must handle the non-stationary data distributions, commonly referred to as concept drift.
arXiv:2608. 19488v1 Announce Type: new Abstract: Production machine learning systems degrade under concept drift, yet practitioners have little principled guidance on when to retrain.
Software-agent benchmarks usually report whether an agent solves a task, but the agent reaches that outcome through a harness that controls what it sees, which actions it can take, which failures are repaired, which states are verified, and which evidence is logged. We show that this harness can change the agent's multi-step beliefs even when the task, environment, and base LLM are fixed.