Under this setting, previous SSL methods tend to predict wrong pseudo-labels with the model fitted on labeled data, resulting in noise accumulation.

Such a problem seriously hinders the realistic landing of classical SSL.

Universal Semi-Supervised Learning (UniSSL) aims to solve the open-set problem where both the class distribution (i.e., class set) and feature distribution (i.e., feature domain) are different between labeled dataset and unlabeled dataset. Such a problem seriously hinders the realistic landing of classical SSL. Different from the existing SSL methods targeting at the open-set problem that only study one certain scenario of class distribution mismatch and ignore the feature distribution mismatch, we consider a more general case where a mismatch exists in both class and feature distribution. In this case, we propose a ''Class-shAring data detection and Feature Adaptation'' (CAFA) framework which requires no prior knowledge of the class relationship between the labeled dataset and unlabeled dataset. Particularly, CAFA utilizes a novel scoring strategy to detect the data in the shared class set. Then, it conducts domain adaptation to fully exploit the value of the detected class-sharing data for better semi-supervised consistency training. Exhaustive experiments on several benchmark datasets show the effectiveness of our method in tackling open-set problems.

Model-based reinforcement learning methods learn a dynamics model with real data sampled from the environment and leverage it to generate simulated data to derive an agent. However, due to the potential distribution mismatch between simulated data and real data, this could lead to degraded performance. Despite much effort being devoted to reducing this distribution mismatch, existing methods fail to solve it explicitly. In this paper, we investigate how to bridge the gap between real and simulated data due to inaccurate model estimation for better policy optimization. To begin with, we first derive a lower bound of the expected return, which naturally inspires a bound maximization algorithm by aligning the simulated and real data distributions. To this end, we propose a novel model-based reinforcement learning framework AMPO, which introduces unsupervised model adaptation to minimize the integral probability metric (IPM) between feature distributions from real and simulated data. Instantiating our framework with Wasserstein-1 distance gives a practical model-based approach. Empirically, our approach achieves state-of-the-art performance in terms of sample efficiency on a range of continuous control benchmark tasks.

Bayesian Optimisation (BO) is a powerful tool for optimising expensive blackbox functions but its effectiveness diminishes in highdimensional spaces due to sparse data and poor surrogate model scalability While Variational Autoencoder (VAE) based approaches address this by learning low-dimensional latent representations the reconstructionbased objective function often brings the functional distribution mismatch between the latent space and original space leading to suboptimal optimisation performance In this paper we first analyse the reason why reconstructiononly loss may lead to distribution mismatch and then propose HiBBO a novel BO framework that introduces the space consistency into the latent space construction in VAE using HiPPO - a method for longterm sequence modelling - to reduce the functional distribution mismatch between the latent space and original space Experiments on highdimensional benchmark tasks demonstrate that HiBBO outperforms existing VAEBO methods in convergence speed and solution quality Our work bridges the gap between high-dimensional sequence representation learning and efficient Bayesian Optimisation enabling broader applications in neural architecture search materials science and beyond.

Under this setting, previous SSL methods tend to predict wrong pseudo-labels with the model fitted on labeled data, resulting in noise accumulation.

Supervised fine-tuning (SFT) of large language models can be viewed as an off-policy learning problem, where expert demonstrations come from a fixed behavior policy while training aims to optimize a target policy. Importance sampling is the standard tool for correcting this distribution mismatch, but large policy gaps lead to skewed weights, high variance, and unstable optimization. Existing methods mitigate this issue with KL penalties or clipping, which passively restrict updates rather than actively reducing the gap. We propose a simple yet effective data rewriting framework that proactively shrinks the policy gap before training. For each problem, correct model-generated solutions are kept as on-policy data, while incorrect ones are rewritten through guided re-solving, falling back to expert demonstrations only when needed. This aligns the training distribution with the target policy, reducing variance and improving stability. To handle residual mismatch after rewriting, we additionally apply importance sampling during training, forming a two-stage approach that combines data-level alignment with lightweight optimization-level correction. Experiments on five mathematical reasoning benchmarks show consistent and significant gains over both vanilla SFT and the state-of-the-art Dynamic Fine-Tuning (DFT) approach. Data and code will be released at https://github.com/NKU-HLT/Off-Policy-SFT.

Moreover, we will conduct additional experiments to further evaluate our method in Section C. Furthermore, we provide the standard deviation results that correspond to the main paper in Section D. Finally, we will discuss the limitations and social impact of our method in Section E. VisDA2017 datasets, we set the batch size to 64. Other implementation details are presented below. It contains 3 domains: "Amazon" (A), "DSLR" (D), and "Webcam" (W), and each domain is composed of 31 classes. Shared learning rate decay factor 0.2 # training iteration in which learning rate decay starts 400,000 # training iteration in which consistency coefficient ramp up starts 200,000 Supervised Initial learning rate 0.003 Π-Model [6, 10] Initial learning rate 3 10 CAFA framework, which includes class-sharing data detection and feature adaptation . Here we use PI as the backbone method.

Such a problem seriously hinders the realistic landing of classical SSL.