Designing and deriving effective model-based reinforcement learning (MBRL) algorithms with a performance improvement guarantee is challenging, mainly attributed to the high coupling between model learning and policy optimization. Many prior methods that rely on return discrepancy to guide model learning ignore the impacts of model shift, which can lead to performance deterioration due to excessive model updates. Other methods use performance difference bound to explicitly consider model shift. However, these methods rely on a fixed threshold to constrain model shift, resulting in a heavy dependence on the threshold and a lack of adaptability during the training process. In this paper, we theoretically derive an optimization objective that can unify model shift and model bias and then formulate a fine-tuning process. This process adaptively adjusts the model updates to get a performance improvement guarantee while avoiding model over-fitting.

We first redefine notation for clarity and then provide the proofs of the results in the main paper. Now we first prove that the iteration in Eq.2 has a fixed point. Proof of Lemma 3.1: Let We present the bound on using empirical Bellman operator compared to the true Bellman operator. The proof can be found in [6]. Proof of Theorem 3.4: Recall that the expression of the V -function iterate is given by: Proof of Theorem 3.6: The proof of this statement is divided into two parts.

Pre-trained vision-language models (e.g., CLIP) have shown powerful zero-shot transfer capabilities. But they still struggle with domain shifts and typically require labeled data to adapt to downstream tasks, which could be costly. In this work, we aim to leverage unlabeled data that naturally spans multiple domains to enhance the transferability of vision-language models. Under this unsupervised multi-domain setting, we have identified inherent model bias within CLIP, notably in its visual and text encoders. Specifically, we observe that CLIP's visual encoder tends to prioritize encoding domain over discriminative category information, meanwhile its text encoder exhibits a preference for domain-relevant classes.

Deep generative networks trained via maximum likelihood on a natural image dataset like CIFAR10 often assign high likelihoods to images from datasets with different objects (e.g., SVHN). We refine previous investigations of this failure at anomaly detection for invertible generative networks and provide a clear explanation of it as a combination of model bias and domain prior: Convolutional networks learn similar low-level feature distributions when trained on any natural image dataset and these low-level features dominate the likelihood. Hence, when the discriminative features between inliers and outliers are on a high-level, e.g., object shapes, anomaly detection becomes particularly challenging. To remove the negative impact of model bias and domain prior on detecting high-level differences, we propose two methods, first, using the log likelihood ratios of two identical models, one trained on the in-distribution data (e.g., CIFAR10) and the other one on a more general distribution of images (e.g., 80 Million Tiny Images). We also derive a novel outlier loss for the in-distribution network on samples from the more general distribution to further improve the performance. Secondly, using a multi-scale model like Glow, we show that low-level features are mainly captured at early scales. Therefore, using only the likelihood contribution of the final scale performs remarkably well for detecting high-level feature differences of the out-of-distribution and the in-distribution. This method is especially useful if one does not have access to a suitable general distribution. Overall, our methods achieve strong anomaly detection performance in the unsupervised setting, and only slightly underperform state-of-the-art classifier-based methods in the supervised setting.

We thank the reviewers for their informative feedback, indicating improved results (All), that hypotheses are "intuitive" (see Section 6). Do partially joint models help? Still, it is interesting future work to try a joint network (see Discussion p.8). That shows local low-level features, beyond being correlated with the likelihood, dominate it. Overclaiming wrt MSP-OE (R5): We agree and would modify wording, e.g., to "slightly underperform".

Air pollution is the world's largest environmental risk factor for human disease and premature death, resulting in more than 6 million permature deaths in 2019. Currently, there is still a challenge to model one of the most important air pollutants, surface ozone, particularly at scales relevant for human health impacts, with the drivers of global ozone trends at these scales largely unknown, limiting the practical use of physics-based models. We employ a 2D Convolutional Neural Network based architecture that estimate surface ozone MOMO-Chem model residuals, referred to as model bias. We demonstrate the potential of this technique in North America and Europe, highlighting its ability better to capture physical model residuals compared to a traditional machine learning method. We assess the impact of incorporating land use information from high-resolution satellite imagery to improve model estimates. Importantly, we discuss how our results can improve our scientific understanding of the factors impacting ozone bias at urban scales that can be used to improve environmental policy.

As large language models (LLMs) are increasingly deployed across diverse linguistic and cultural contexts, understanding their behavior in both factual and disputable scenarios is essential, especially when their outputs may shape public opinion or reinforce dominant narratives. In this paper, we define two types of bias in LLMs: model bias (bias stemming from model training) and inference bias (bias induced by the language of the query), through a two-phase evaluation. Phase 1 evaluates LLMs on factual questions where a single verifiable answer exists, assessing whether models maintain consistency across different query languages. Phase 2 expands the scope by probing geopolitically sensitive disputes, where responses may reflect culturally embedded or ideologically aligned perspectives. We construct a manually curated dataset spanning both factual and disputable QA, across four languages and question types. The results show that Phase 1 exhibits query language induced alignment, while Phase 2 reflects an interplay between the model's training context and query language. This paper offers a structured framework for evaluating LLM behavior across neutral and sensitive topics, providing insights for future LLM deployment and culturally aware evaluation practices in multilingual contexts.