Can we train a 3D molecule generator using data from dense regions to generate samples in sparse regions? This challenge can be framed as an out-of-distribution (OOD) generation problem. While prior research on OOD generation predominantly targets property shifts, structural shifts, such as differences in molecular scaffolds or functional groups, represent an equally critical source of distributional shifts. This work introduces the Geometric OOD Diffusion Model (GODD), a novel diffusion-based framework that enables training on data-abundant molecular distributions while generalizing to data-scarce distributions under distributional structural shifts. Central to our approach is a designated equivariant asymmetric autoencoder to capture distributional structural priors. The asymmetric design allows the model to generalize to unseen structural variations by capturing distributional priors representing distinct distributions. The encoded structural-grained priors guide generation toward sparse regions without requiring explicit training on such data. Evaluated across standard benchmarks encompassing OOD structural shifts (e.g., scaffolds, rings), GODD achieves an improvement of 12.6% in success rate, defined based on molecular validity, uniqueness, and novelty. Furthermore, the framework demonstrates promising performance and generalization on canonical fragment-based drug design tasks, highlighting its utility in learning-based molecular discovery.

The widespread adoption of Large Language Models (LLMs) has raised significant privacy concerns regarding the exposure of personally identifiable information (PII) in user prompts.To address this challenge, we propose a queryunrelated PII masking strategy and introduce PII-Bench, the first comprehensive evaluation framework for assessing privacy protection systems.PII-Bench comprises 2,842 test samples across 7 PII types with 55 fine-grained subcategories, featuring diverse scenarios from singlesubject descriptions to complex multi-party interactions.Each sample is carefully crafted with a user query, context description, and standard answer indicating query-relevant PII.Our empirical evaluation reveals that while current models perform adequately in basic PII detection, they show significant limitations in determining PII query relevance.Even advanced LLMs struggle with this task, particularly in handling complex multi-subject scenarios, indicating substantial room for improvement in achieving intelligent PII masking.

The design of new crystal materials is of significant scientific importance to society. In recent years, machine learning-based approaches have shown their potential in crystal material design. However, their effectiveness relies heavily on the availability of high-quality and extensive training data, which is difficult to collect in practice. To this end, this paper presents a novel physics-informed evolutionary transfer optimization framework that can design new crystal materials without the need for extensive data. Specifically, we first propose a novel physics-informed encoding for materials, enabling the use of multi-objective evolutionary optimization to simultaneously optimize multiple physical objectives, including the validity, properties, and energy of crystal materials. These physical objectives are critical to the effective design of crystal materials. Additionally, to mitigate the slow optimization speed of evolutionary computation, we propose a physics-informed evolutionary transfer optimization technique to enhance the design speed of optimized materials. We conducted comprehensive experiments to analyze the designed crystals from the perspectives of validity, density functional theory (DFT) validation, formation energy, and energy above hull. The experimental results validate the immense potential of the proposed physics-informed multi-objective evolutionary optimization framework in crystal material design.

This paper proposes a new 3D molecule generation framework, called GOAT, for fast and effective 3D molecule generation based on the flow-matching optimal transport objective. Specifically, we formulate a geometric transport formula for measuring the cost of mapping multi-modal features (e.g., continuous atom coordinates and categorical atom types) between a base distribution and a target data distribution. Our formula is solved within a joint, equivariant, and smooth representation space. This is achieved by transforming the multi-modal features into a continuous latent space with equivariant networks. In addition, we find that identifying optimal distributional coupling is necessary for fast and effective transport between any two distributions. We further propose a mechanism for estimating and purifying optimal coupling to train the flow model with optimal transport. By doing so, GOAT can turn arbitrary distribution couplings into new deterministic couplings, leading to an estimated optimal transport plan for fast 3D molecule generation. The purification filters out the subpar molecules to ensure the ultimate generation quality. We theoretically and empirically prove that the proposed optimal coupling estimation and purification yield transport plan with non-increasing cost. Finally, extensive experiments show that GOAT enjoys the efficiency of solving geometric optimal transport, leading to a double speedup compared to the sub-optimal method while achieving the best generation quality regarding validity, uniqueness, and novelty.

Neural operators are a class of neural networks to learn mappings between infinite-dimensional function spaces, and recent studies have shown that using neural operators to solve partial differential equations is a promising direction. However, the latest neural operator-based algorithms still leave space for increasing the running speed of algorithms while maintaining accuracy. In this work, we propose a new neural operator by using a neural network to "imitate" the process of the Fourier Transform to solve partial differential equations, called the Imitative Fourier Neural Operator (IFNO). The advantage of this method is to use the property of Fourier Transform to transform the partial differential equation into an algebraic equation, which reduces the fitting difficulty of the neural network and improves the training speed. We compare the proposed algorithm with several state-of-the-art designs on different benchmark instances. The experimental results confirm the effectiveness and performance of the proposed method for solving partial differential equations.

The large-scale multiobjective optimization problem (LSMOP) is characterized by simultaneously optimizing multiple conflicting objectives and involving hundreds of decision variables. Many real-world applications in engineering can be modeled as LSMOPs; simultaneously, engineering applications require insensitivity in performance. This requirement typically means that the algorithm should not only produce good results in terms of performance for every run but also the performance of multiple runs should not fluctuate too much. However, existing large-scale multiobjective optimization algorithms often focus on improving algorithm performance, but pay little attention to improving the insensitivity characteristic of algorithms. This directly leads to substantial limitations when solving practical problems. In this work, we propose an evolutionary algorithm called large-scale multiobjective optimization algorithm via Monte Carlo tree search, which is based on the Monte Carlo tree search and aims to improve the performance and insensitivity of solving LSMOPs. The proposed method samples decision variables to construct new nodes on the Monte Carlo tree for optimization and evaluation, and it selects nodes with good evaluations for further searches in order to reduce the performance sensitivity caused by large-scale decision variables. We propose two metrics to measure the sensitivity of the algorithm and compare the proposed algorithm with several state-of-the-art designs on different benchmark functions and metrics. The experimental results confirm the effectiveness and performance insensitivity of the proposed design for solving LSMOPs.

Large-scale multiobjective optimization problems (LSMOPs) are characterized as optimization problems involving hundreds or even thousands of decision variables and multiple conflicting objectives. To solve LSMOPs, some algorithms designed a variety of strategies to track Pareto-optimal solutions (POSs) by assuming that the distribution of POSs follows a low-dimensional manifold. However, traditional genetic operators for solving LSMOPs have some deficiencies in dealing with the manifold, which often results in poor diversity, local optima, and inefficient searches. In this work, a generative adversarial network (GAN)-based manifold interpolation framework is proposed to learn the manifold and generate high-quality solutions on the manifold, thereby improving the optimization performance of evolutionary algorithms. We compare the proposed approach with several state-of-the-art algorithms on various large-scale multiobjective benchmark functions. The experimental results demonstrate that significant improvements have been achieved by the proposed framework in solving LSMOPs.

Large-scale multiobjective optimization problems (LSMOPs) refer to optimization problems with multiple con-flicting optimization objectives and hundreds or even thousands of decision variables. A key point in solving LSMOPs is how to balance exploration and exploitation so that the algorithm can search in a huge decision space efficiently. Large-scale multi-objective evolutionary algorithms consider the balance between exploration and exploitation from the individual's perspective. However, these algorithms ignore the significance of tackling this issue from the perspective of decision variables, which makes the algorithm lack the ability to search from different dimensions and limits the performance of the algorithm. In this paper, we propose a large-scale multiobjective optimization algorithm based on the attention mechanism, called (LMOAM). The attention mechanism will assign a unique weight to each decision variable, and LMOAM will use this weight to strike a balance between exploration and exploitation from the decision variable level. Nine different sets of LSMOP benchmarks are conducted to verify the algorithm proposed in this paper, and the experimental results validate the effectiveness of our design.

Dynamic multiobjective optimization problems (DMOPs) are optimization problems with multiple conflicting optimization objectives, and these objectives change over time. Transfer learning-based approaches have been proven to be promising; however, a slow solving speed is one of the main obstacles preventing such methods from solving real-world problems. One of the reasons for the slow running speed is that low-quality individuals occupy a large amount of computing resources, and these individuals may lead to negative transfer. Combining high-quality individuals, such as knee points, with transfer learning is a feasible solution to this problem. However, the problem with this idea is that the number of high-quality individuals is often very small, so it is difficult to acquire substantial improvements using conventional transfer learning methods. In this article, we propose a knee point-based transfer learning method, called KT-DMOEA, for solving DMOPs. In the proposed method, a trend prediction model (TPM) is developed for producing the estimated knee points. Then, an imbalance transfer learning method is proposed to generate a high-quality initial population by using these estimated knee points. The advantage of this approach is that the seamless integration of a small number of high-quality individuals and the imbalance transfer learning technique can greatly improve the computational efficiency while maintaining the quality of the solution. The experimental results and performance comparisons with some chosen state-of-the-art algorithms demonstrate that the proposed design is capable of significantly improving the performance of dynamic optimization.