session-adaptive incremental training

This paper addresses five core challenges in dynamic data environments—data drift, concept drift, catastrophic forgetting, skewed learning, and network adaptability. Method: It systematically surveys over 120 state-of-the-art evolutionary machine learning (EML) works, integrating online learning, incremental learning, continual learning, meta-learning, dynamic pruning, and ensemble distillation to establish a multi-paradigm evaluation framework covering supervised, unsupervised, and semi-supervised settings. Contribution/Results: The work introduces the first unified analytical framework for EML, clarifies challenge taxonomies, uncovers synergistic mechanisms among adaptive neural architectures, meta-learning, and ensemble strategies, and identifies critical gaps in robustness, ethics, and scalability. It delivers a comprehensive EML methodology landscape, a curated collection of mainstream benchmarks and evaluation metrics, and system design principles tailored for industrial deployment—providing both theoretical foundations and practical guidance for building dynamic AI systems.

Traditional continual learning is constrained by a parameter-centric paradigm, limiting its capacity to meet system-level adaptation demands in dynamic environments. This work proposes a “Tri-Axis Framework” (When, How, Where), offering a unified perspective that reorients continual learning beyond mere parameter updates toward external architectures and inference-time adaptation. By integrating off-policy/on-policy learning, test-time training, external memory systems, and skill repositories, the framework transcends the limitations of static parameter spaces and gradient-based optimization. A systematic review elucidates the field’s evolutionary trajectory and highlights pivotal challenges and future directions inherent in this paradigm shift.

This work addresses the lack of a general, auditable dynamic control mechanism in existing training systems, which typically rely on framework-specific code. The authors propose the first cross-framework, open-source control plane that exposes training interfaces through a unified protocol, integrating declarative configuration, request validation, and secure control-point scheduling within the Aim workspace to enable metric monitoring, real-time intervention, and operational traceability. The system supports safe human and automated controller interventions during training while fully logging all operational trajectories. Experiments across five NLP and reinforcement learning tasks demonstrate its effectiveness, and the open-source implementation provides a foundation for reproducible human-in-the-loop training.

In continual learning, artificial neural networks suffer from catastrophic forgetting—performance on previously learned tasks degrades significantly upon training on new tasks. Existing approaches rely on heuristic task-scheduling protocols lacking theoretical guarantees of optimality. This paper bridges statistical physics and optimal control theory to establish, for the first time, an analytically tractable and provably optimal framework for task selection dynamics. Leveraging a teacher–student model, we derive exact training dynamics via dynamic mean-field analysis and obtain a closed-form optimal scheduling protocol that explicitly incorporates task similarity as a key regulator of forgetting. Empirical evaluation on synthetic data and real-world benchmarks (e.g., CIFAR-100) demonstrates substantial reduction in forgetting rates. Crucially, theoretical predictions align closely with experimental results, validating the framework’s strong interpretability, formal optimality guarantee, and cross-dataset generalizability.

Current continual learning (CL) evaluation protocols suffer from critical flaws—hyperparameter tuning and evaluation are conducted within the same scenario, leading to systematic overestimation of CL capability and employing unrealistic, non-deployable tuning practices. Method: We propose the Generalized Two-stage Evaluation Protocol (GTEP), which strictly decouples hyperparameter optimization (performed solely on a source dataset) from performance evaluation (conducted on a target dataset), thereby enforcing cross-dataset generalization under structurally identical tasks. Contribution/Results: Extensive experiments—over 8,000 runs across CIFAR and ImageNet variants—under both pre-trained and non-pre-trained settings within a class-incremental learning framework demonstrate that mainstream SOTA methods suffer 30–50% average performance degradation under GTEP. This reveals their lack of robustness across deployment scenarios and establishes GTEP as a more rigorous, realistic benchmark for trustworthy continual learning.

Existing approaches to incremental learning for large language models (LLMs) suffer from high catastrophic forgetting, poor generalization, or inability to support true online adaptation—largely because they fail to enable real-time, progressive updates to LLMs’ core parameters. Method: We systematically survey four major paradigms—continual learning, meta-learning, parameter-efficient fine-tuning (e.g., LoRA, Adapters), and mixture-of-experts (MoE)—and identify, for the first time, that none achieve genuine real-time incremental updates to LLMs’ fundamental weight matrices. Contribution/Results: Based on this analysis, we propose the first comprehensive taxonomy for LLM incremental learning, rigorously delineating the capabilities and applicability boundaries of each paradigm. Furthermore, we articulate a forward-looking research direction that jointly prioritizes low catastrophic forgetting, strong cross-task knowledge generalization, and seamless online parameter updateability—establishing foundational principles for next-generation adaptive LLMs.

This work addresses the fragmented landscape of post-training adaptation techniques, which suffer from inconsistent terminology and a lack of unified comparative or governance frameworks. To resolve this, the paper introduces the first six-dimensional taxonomy—spanning mechanism, objective, data requirements, persistence, structural scope, and model type—that systematically integrates mainstream approaches such as fine-tuning, retrieval augmentation, prompt engineering, model editing, and machine unlearning. This framework clarifies conceptual boundaries and reveals evolutionary and compositional relationships among methods. Beyond standardizing terminology, it enables standardized technical documentation, model change tracking, and AI governance analysis. The study further identifies critical challenges, including evaluation rigor, reproducibility, continual adaptation, multimodal alignment, and governance-aware workflows.

This work addresses catastrophic forgetting in continual learning under non-stationary data streams by proposing the COLD framework, which introduces, for the first time, the Drift-Plus-Penalty stochastic optimization method from control theory into this domain. COLD formulates forgetting as a controlled dynamic process, employing virtual queues to track performance deviations on historical tasks and jointly minimizing the current task loss and queue drift at each optimization step. This mechanism explicitly governs the stability-plasticity trade-off. The framework provides theoretical guarantees on stability and convergence, and achieves significantly superior performance over state-of-the-art methods on standard benchmarks, enabling controllable and efficient suppression of catastrophic forgetting.

Existing time series pretraining methods struggle to generalize effectively across multiple datasets due to discrepancies in input length and channel dimensions. This work proposes ADAPT, a novel pretraining paradigm that enables unified modeling across 162 time series classification datasets by adaptively aligning the physical attributes of time series data. Integrating self-supervised learning with a hybrid batch training strategy, ADAPT overcomes the generalization limitations inherent in conventional many-to-one pretraining approaches. The method achieves state-of-the-art performance on multiple benchmarks, establishing a foundational framework for developing general-purpose foundation models for time series analysis.