Large Language Models (LLMs) demonstrate outstanding performance in their reservoir of knowledge and understanding capabilities, but they have also been shown to be prone to illegal or unethical reactions when subjected to jailbreak attacks. To ensure their responsible deployment in critical applications, it is crucial to understand the safety capabilities and vulnerabilities of LLMs. Previous works mainly focus on jailbreak in single-round dialogue, overlooking the potential jailbreak risks in multi-round dialogues, which are a vital way humans interact with and extract information from LLMs. Some studies have increasingly concentrated on the risks associated with jailbreak in multi-round dialogues. These efforts typically involve the use of manually crafted templates or prompt engineering techniques. However, due to the inherent complexity of multi-round dialogues, their jailbreak performance is limited. To solve this problem, we propose a novel multi-round dialogue jailbreaking agent, emphasizing the importance of stealthiness in identifying and mitigating potential threats to human values posed by LLMs. We propose a risk decomposition strategy that distributes risks across multiple rounds of queries and utilizes psychological strategies to enhance attack strength. Extensive experiments show that our proposed method surpasses other attack methods and achieves state-of-the-art attack success rate. We will make the corresponding code and dataset available for future research. The code will be released soon.

The application of large deep learning models in weather forecasting has led to significant advancements in the field, including higher-resolution forecasting and extended prediction periods exemplified by models such as Pangu and Fuxi. Despite these successes, previous research has largely been characterized by the neglect of extreme weather events, and the availability of datasets specifically curated for such events remains limited. Given the critical importance of accurately forecasting extreme weather, this study introduces a comprehensive dataset that incorporates high-resolution extreme weather cases derived from the High-Resolution Rapid Refresh (HRRR) data, a 3-km real-time dataset provided by NOAA. We also evaluate the current state-of-the-art deep learning models and Numerical Weather Prediction (NWP) systems on HR-Extreme, and provide a improved baseline deep learning model called HR-Heim which has superior performance on both general loss and HR-Extreme compared to others. Our results reveal that the errors of extreme weather cases are significantly larger than overall forecast error, highlighting them as an crucial source of loss in weather prediction. These findings underscore the necessity for future research to focus on improving the accuracy of extreme weather forecasts to enhance their practical utility.

Federated learning is a contemporary machine learning paradigm where locally trained models are distilled into a global model. Due to the intrinsic permutation invariance of neural networks, Probabilistic Federated Neural Matching (PFNM) employs a Bayesian nonparametric framework in the generation process of local neurons, and then creates a linear sum assignment formulation in each alternative optimization iteration. But according to our theoretical analysis, the optimization iteration in PFNM omits global information from existing. In this study, we propose a novel approach that overcomes this flaw by introducing a Kullback-Leibler divergence penalty at each iteration. The effectiveness of our approach is demonstrated by experiments on both image classification and semantic segmentation tasks.

Abstract--In federated learning, models trained on local clients are distilled into a global model. Due to the permutation invariance arises in neural networks, it is necessary to match the hidden neurons first when executing federated learning with neural networks. Through the Bayesian nonparametric framework, Probabilistic Federated Neural Matching (PFNM) matches and fuses local neural networks so as to adapt to varying global model size and the heterogeneity of the data. In this paper, we propose a new method which extends the PFNM with a Kullback-Leibler (KL) divergence over neural components product, in order to make inference exploiting posterior information in both local and global levels. We also show theoretically that The additional part can be seamlessly concatenated into the match-and-fuse progress. Through a series of simulations, it indicates that our new method outperforms popular state-of-the-art federated learning methods in both single communication round and additional communication rounds situation. Considering the inherently private attribute of the data and communication cost, pooling data from many devices at the data center, and conducting centralized training is far from ideal. Thus, federated learning (FL), in which a model is learned from siloed data without centralized together, is proposed to tackle this challenge.

Energy disaggregation in a non-intrusive way estimates appliance level electricity consumption from a single meter that measures the whole house electricity demand. Recently, with the ongoing increment of energy data, there are many data-driven deep learning architectures being applied to solve the non-intrusive energy disaggregation problem. However, most proposed methods try to estimate the on-off state or the power consumption of appliance, which need not only large amount of parameters, but also hyper-parameter optimization prior to training and even preprocessing of energy data for a specified appliance. In this paper, instead of estimating on-off state or power consumption, we adapt a neural network to estimate the operational state change of appliance. Our proposed solution is more feasible across various appliances and lower complexity comparing to previous methods. The simulated experiments in the low sample rate dataset REDD show the competitive performance of the designed method, with respect to other two benchmark methods, Hidden Markov Model-based and Graph Signal processing-based approaches.

Classical inconsistency-tolerant query answering relies on selecting maximal components of an ABox/database which are consistent with the ontology. However, some rules in ontologies might be unreliable if they are extracted from ontology learning or written by unskillful knowledge engineers. In this paper we present a framework of handling inconsistent existential rules under stable model semantics, which is defined by a notion called rule repairs to select maximal components of the existential rules. Surprisingly, for R-acyclic existential rules with R-stratified or guarded existential rules with stratified negations, both the data complexity and combined complexity of query answering under the rule repair semantics remain the same as that under the conventional query answering semantics. This leads us to propose several approaches to handle the rule repair semantics by calling answer set programming solvers. An experimental evaluation shows that these approaches have good scalability of query answering under rule repairs on realistic cases.

This paper proposes an alternative definition of elementary loops and extends the notion of proper loops for disjunctive logic programs. Different from normal logic programs, the computational complexities of recognizing elementary loops and proper loops for disjunctive programs are coNP-complete. To address this problem, we introduce weaker versions of both elementary loops and proper loops and provide polynomial time algorithms for identifying them respectively. On the other hand, based on the notion of elementary loops, the class of Head-Elementary-loop-Free (HEF) programs was presented, which can be turned into equivalent normal logic programs by shifting head atoms into bodies. However, the problem of recognizing an HEF program is coNP-complete. Then we present a subclass of HEF programs which generalizes the class of Head-Cycle-Free programs and provide a polynomial time algorithm to identify them. At last, some experiments show that both elementary loops and proper loops could be replaced by their weak versions in practice.

The notions of loops and loop formulas play an important role in answer set computation. However, there would be an exponential number of loops in the worst case. Gebser and Schaub characterized a subclass elementary loops and showed that they are sufficient for selecting answer sets from models of a logic program. This paper proposes an alternative definition of elementary loops and identify a subclass of elementary loops, called proper loops. By applying a special form of their loop formulas, proper loops are also sufficient for the SAT-based answer set computation. A polynomial algorithm to recognize a proper loop is given and shows that for certain logic programs, identifying all proper loops of a program is more efficient than that of elementary loops. Furthermore, we prove that, by considering the structure of the positive body-head dependency graph of a program, a large number of loops could be ignored for identifying proper loops. We provide another algorithm for identifying all proper loops of a program. The experiments show that, for certain programs whose dependency graphs consisting of sets of components that are densely connected inside and sparsely connected outside, the new algorithm is more efficient.