Yifei Zhang, Bo Pan, Chen Ling, Yuntong Hu, Liang Zhao. Findings of the Association for Computational Linguistics ACL 2024. 2024.

The Split Hopkinson Pressure Bar (SHPB) is an apparatus for testing the dynamic stress-strain response of the cement mortar specimen with pre-set joints at different angles to explore the influence of joint attitudes of underground rock engineering on the failure characteristics of rock mass structure. The nuclear magnetic resonance (NMR) has also been used to measure the pore distribution and internal cracks of the specimen before and after the testing. In combination with numerical analysis, the paper systematically discusses the influence of joint angles on the failure mode of rock-like materials from three aspects of energy dissipation, microscopic damage, and stress field characteristics. The result indicates that the impact energy structure of the SHPB is greatly affected by the pre-set joint angle of the specimen. With the joint angle increasing, the proportion of reflected energy moves in fluctuation, while the ratio of transmitted energy to dissipated energy varies from one to the other. NMR analysis reveals the structural variation of the pores in those cement specimens before and after the impact. Crack propagation direction is correlated with pre-set joint angles of the specimens. With the increase of the pre-set joint angles, the crack initiation angle decreases gradually. When the joint angles are around 30°–75°, the specimens develop obvious cracks. The crushing process of the specimens is simulated by LS-DYNA software. It is concluded that the stresses at the crack initiation time are concentrated between 20 and 40 MPa. The instantaneous stress curve first increases and then decreases with crack propagation, peaking at different times under various joint angles; but most of them occur when the crack penetration ratio reaches 80–90%. With the increment of joint angles in specimens through the simulation software, the changing trend of peak stress is consistent with the test results.

Classical Federated Learning (FL) encounters significant challenges when deploying large models on power-constrained clients. To tackle this, we propose an asymmetric FL mechanism that enables the aggregation of compact client models into a comprehensive server model. We design the server model as a Mixture-of-Experts (MoE), where each expert has the same architecture as each client model. This uniformity allows for efficient fusion of the most pertinent client models to update each server expert, based on the measured relevance between each client and server expert. To address the Non-IID data issue, we further optimize the server-side MoE architecture by incorporating a main expert that always activates alongside a set of selectively activated routed experts. This configuration ensures a balance between learning general knowledge and specific data distribution. Our Fed-MoE framework is model-agnostic and has demonstrated notable improvements on vision FL tasks with million-scale ResNet backbones, and language tasks with billion-scale BERT and GPT-2 backbones.

Developing deep generative models has been an emerging field due to the ability to model and generate complex data for various purposes, such as image synthesis and molecular design. However, the advancement of deep generative models is limited by challenges to generate objects that possess multiple desired properties: 1) the existence of complex correlation among real-world properties is common but hard to identify; 2) controlling individual property enforces an implicit partially control of its correlated properties, which is difficult to model; 3) controlling multiple properties under various manners simultaneously is hard and under-explored. We address these challenges by proposing a novel deep generative framework that recovers semantics and the correlation of properties through disentangled latent vectors. The correlation is handled via an explainable mask pooling layer, and properties are precisely retained by generated objects via the mutual dependence between latent vectors and properties. Our generative model preserves properties of interest while handling correlation and conflicts of properties under a multi-objective optimization framework. The experiments demonstrate our model's superior performance in generating data with desired properties.

Matched molecular pairs (MMPs) capture the local chemical edits that medicinal chemists routinely use to design analogs, but existing ML approaches either operate at the whole-molecule level with limited edit controllability or learn MMP-style edits from restricted settings and small models. We propose a variable-to-variable formulation of analog generation and train a foundation model on large-scale MMP transformations (MMPTs) to generate diverse variables conditioned on an input variable. To enable practical control, we develop prompting mechanisms that let the users specify preferred transformation patterns during generation. We further introduce MMPT-RAG, a retrieval-augmented framework that uses external reference analogs as contextual guidance to steer generation and generalize from project-specific series. Experiments on general chemical corpora and patent-specific datasets demonstrate improved diversity, novelty, and controllability, and show that our method recovers realistic analog structures in practical discovery scenarios.

Visual explanation (attention)-guided learning uses not only labels but also explanations to guide model reasoning process. While visual attention-guided learning has shown promising results, it requires a large number of explanation annotations that are time-consuming to prepare. However, in many real-world situations, it is usually desired to prompt the model with visual attention without model retraining. For example, when doing AI-assisted cancer classification on a medical image, users (e.g., clinicians) can provide the AI model with visual attention prompt on which areas are indispensable and which are precluded. Despite its promising objectives, achieving visual attention-prompted prediction presents several major challenges: 1) How can the visual prompt be effectively integrated into the model's reasoning process? 2) How should the model handle samples that lack visual prompts? 3) What is the impact on the model's performance when a visual prompt is imperfect? This paper introduces a novel framework for attention-prompted prediction and learning, utilizing visual prompts to steer the model's reasoning process. To improve performance in non-prompted situations and align it with prompted scenarios, we propose a co-training approach for both non-prompted and prompted models, ensuring they share similar parameters and activations. Additionally, for instances where the visual prompt does not encompass the entire input image, we have developed innovative attention prompt refinement methods. These methods interpolate the incomplete prompts while maintaining alignment with the model's explanations. Extensive experiments on four datasets demonstrate the effectiveness of our proposed framework in enhancing predictions for samples both with and without prompt.