collaborative source-free adaptation

This paper addresses unsupervised model adaptation from a source domain to a target domain without access to any source-domain data or labels—termed *source-free* adaptation—aiming solely to improve the generalization of a pre-trained source model on unlabeled target data. To this end, we propose a Collaborative Class-Conditional Generative Adversarial Network (CC-GAN) framework: it models the semantic structure of the target domain via class-conditional generation; enforces weight constraints derived from the source model to preserve its discriminative capability; and incorporates clustering-driven feature regularization to enhance the discriminability of target-domain representations. Evaluated across multiple cross-domain vision tasks, our method achieves significant performance gains over conventional source-dependent adaptation approaches—using only unlabeled target data. It is the first to empirically validate the effectiveness, robustness, and scalability of model adaptation under the source-free setting.

This work addresses the privacy risks in source-free domain adaptation, where models trained on the source domain may inadvertently leak information about source-private classes into the target domain. To tackle this issue, we formally introduce and solve the problem of source-private class forgetting under a novel setting termed SCADA-UL, extending it to scenarios involving continual forgetting and unknown forgotten classes. By leveraging adversarially generated forgetting samples, a label rescaling strategy, and adversarial optimization, our method achieves effective machine unlearning under distribution shift. Experimental results demonstrate that the proposed approach attains forgetting performance comparable to full retraining across multiple benchmark datasets, significantly outperforming existing baselines.

To address the challenges of fine-grained knowledge extraction, inefficient aggregation, and suboptimal accuracy in multi-source transfer learning, this paper proposes a lightweight knowledge fusion framework based on Singular Value Decomposition (SVD). Methodologically, each source model is decomposed layer-wise into rank-one components; salient components are selected according to significance, and only the principal singular values of the fused matrix are fine-tuned for target-task adaptation. This mechanism achieves both high accuracy and efficiency: it avoids full-parameter fine-tuning, drastically reducing retraining overhead; exhibits robustness to noise and pruning-induced perturbations; and scales effectively to large, high-parameter models. Experiments demonstrate substantial performance gains across diverse multi-source transfer tasks, strong computational scalability, and establish a novel paradigm for efficient model knowledge reuse.

This work addresses the challenges of source knowledge forgetting and overfitting to local noise in unsupervised domain adaptation, which often arise from an overreliance on neighborhood prediction similarity. To mitigate these issues, the authors propose a dynamic probability calibration method that leverages a dual-model co-prediction mechanism to adaptively calibrate neighborhood probabilities by fusing the initial outputs of the source model with the online predictions of the current model. The approach jointly optimizes a soft supervision loss and a diversity loss, effectively preserving discriminative source-domain information while suppressing interference from local noise. Extensive experiments across 31 cross-domain tasks on four public datasets demonstrate that the proposed method significantly alleviates knowledge forgetting and overfitting, achieving a balanced integration of source knowledge and target-specific information.

Existing data-free meta-learning methods are constrained to parameter-space optimization and require homogeneous model architectures, limiting scalability to large-scale pretrained models. This paper introduces the first data-free meta-learning framework tailored for heterogeneous pretrained models, enabling extraction of implicit prior knowledge without access to original training data. Our approach features two core innovations: Episode Curriculum Inversion (ECI) and Inversion Calibration Following Inner Loop (ICFIL). Leveraging pseudo-task distillation, adversarial end-to-end meta-training, and curriculum-based pseudo-episode generation, the framework achieves generalizable meta-adaptation across architectural heterogeneity, model scales (up to 10B parameters), and diverse datasets. Experiments demonstrate substantial improvements over state-of-the-art data-free meta-learning methods across multiple benchmarks, validating both strong generalization and seamless scalability.

This paper addresses machine unlearning—the efficient removal of a model’s dependence on specific training samples while preserving performance on the remaining data. We propose a constraint-optimization-based feasible update framework. Our core innovation introduces a parameter masking mechanism to select an updateable subspace, jointly incorporating gradient noise modeling and directional constraints on parameter updates to yield locally feasible solutions satisfying both unlearning objectives and utility preservation. The method operates as a plug-and-play module, enhancing the robustness and accuracy of diverse first-order approximate unlearning algorithms. Experiments on image classification tasks demonstrate that our approach significantly improves unlearning accuracy (average gain of 12.3%) while incurring negligible utility loss—less than 0.5% drop in test accuracy on retained data—validating its effectiveness and practicality.

This work addresses the lack of systematic methodologies in model optimization, which often relies on heuristic choices and struggles to accommodate diverse deployment constraints. It formalizes model compression and acceleration as a constraint-aware multi-objective engineering decision problem, establishing a unified and actionable framework grounded in five key dimensions: data availability, latency, memory footprint, accuracy tolerance, and retraining budget. By integrating techniques such as quantization, pruning, knowledge distillation, parameter-efficient fine-tuning (PEFT), and inference optimization, the study proposes tailored optimization pipelines for four representative industrial scenarios, delivering a reproducible and quantifiable guide for technology selection.

This work proposes a source-free unsupervised domain adaptation method that operates without access to source domain data. By optimizing the intra-batch cosine similarity and dissimilarity of target-domain samples and incorporating a neighborhood signature–based clustering mechanism, the approach effectively suppresses interference from noisy neighbors and constructs more discriminative cluster structures. Notably, the method relies solely on a single loss term, substantially simplifying the adaptation process. Extensive experiments on benchmark datasets—including VisDA—demonstrate its state-of-the-art performance, with particularly significant gains on VisDA that underscore both its effectiveness and novelty.

This work addresses the challenge of targeted machine unlearning in large language models (LLMs). Unlike conventional fine-tuning or data-deletion approaches, we propose an efficient unlearning method grounded in the model editing paradigm. We are the first to systematically evaluate and adapt causal mediation–based editing algorithms—including ROME, IKE, and WISE—for machine unlearning tasks. Crucially, we reformulate the editing objective to emphasize precise knowledge localization and controllable, localized parameter modification—driven by gradients or activations—enabling accurate removal of targeted information. On multiple standard unlearning benchmarks, our method substantially reduces residual memory rates while limiting downstream task performance degradation to under 3%; in certain settings, it outperforms state-of-the-art unlearning baselines. Our core contributions are: (i) establishing model editing as a novel, high-fidelity paradigm for targeted unlearning; and (ii) introducing principled design criteria and technical pathways for unlearning-aware editing objectives.

This work proposes the DMM framework to address scenarios where data privacy or heterogeneity precludes centralized training, enabling effective fusion of highly heterogeneous domain models without accessing raw data. The approach follows a three-stage pipeline: first, individual domain models are trained independently; second, similar models are clustered and merged; third, normalized statistics are leveraged to synthesize pseudo-data for lightweight knowledge distillation of the aggregated model. DMM is the first method to achieve stable model fusion under strict no-data conditions while preserving rare yet critical knowledge through pseudo-data–guided distillation. Experiments demonstrate that DMM consistently outperforms existing model fusion techniques on both single-modal and multimodal benchmarks.