Yuhao Wang, Keyan Ding, Kehua Feng, Zeyuan Wang, Ming Qin, Xiaotong Li, Qiang Zhang, Huajun Chen. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Zeyuan Wang, Qiang Zhang, Keyan Ding, Ming Qin, Xiang Zhuang, Xiaotong Li, Huajun Chen. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Directed evolution, a cornerstone of protein optimization, is to harness natural mutational processes to enhance protein functionality. Existing Machine Learning-assisted Directed Evolution (MLDE) methodologies typically rely on data-driven strategies and often overlook the profound domain knowledge in biochemical fields. In this paper, we introduce a novel Knowledge-aware Reinforced Language Model (KnowRLM) for MLDE. An Amino Acid Knowledge Graph (AAKG) is constructed to represent the intricate biochemical relationships among amino acids. We further propose a Protein Language Model (PLM)-based policy network that iteratively samples mutants through preferential random walks on the AAKG using a dynamic sliding window mechanism. The novel mutants are actively sampled to fine-tune a fitness predictor as the reward model, providing feedback to the knowledge-aware policy. Finally, we optimize the whole system in an active learning approach that mimics biological settings in practice.KnowRLM stands out for its ability to utilize contextual amino acid information from knowledge graphs, thus attaining advantages from both statistical patterns of protein sequences and biochemical properties of amino acids.Extensive experiments demonstrate the superior performance of KnowRLM in more efficiently identifying high-fitness mutants compared to existing methods.

Protein optimization is a fundamental biological task aimed at enhancing theperformance of proteins by modifying their sequences. Computational methodsprimarily rely on evolutionary information (EI) encoded by protein languagemodels (PLMs) to predict fitness landscape for optimization. However, thesemethods suffer from a few limitations. (1) Evolutionary processes involve thesimultaneous consideration of multiple functional properties, often overshadowingthe specific property of interest. (2) Measurements of these properties tend to betailored to experimental conditions, leading to reduced generalizability of trainedmodels to novel proteins. To address these limitations, we introduce DenoisingProtein Language Models (DePLM), a novel approach that refines the evolutionaryinformation embodied in PLMs for improved protein optimization. Specifically, weconceptualize EI as comprising both property-relevant and irrelevant information,with the latter acting as “noise” for the optimization task at hand. Our approachinvolves denoising this EI in PLMs through a diffusion process conducted in therank space of property values, thereby enhancing model generalization and ensuringdataset-agnostic learning. Extensive experimental results have demonstrated thatDePLM not only surpasses the state-of-the-art in mutation effect prediction butalso exhibits strong generalization capabilities for novel proteins.

Directed evolution is a widely-used strategy of protein engineering to improve protein function via mimicking natural mutation and selection. Machine learning-assisted directed evolution (MLDE) approaches aim to learn a fitness predictor, thereby efficiently searching for optimal mutants within the vast combinatorial mutation space. Since annotating mutants is both costly and labor-intensive, how to efficiently sample and utilize informative protein mutants to train the predictor is a critical problem in MLDE. Previous MLDE works just simply utilized pre-trained protein language models (PPLMs) for sampling without tailoring to the specific target protein of interest, which has not fully exploited the potential of PPLMs. In this work, we propose a novel method, the Actively-Finetuned Protein language model for Directed Evolution(AFP-DE), which leverages PPLMs to actively sample and fine-tune themselves, continuously improving the model’s sampling and overall performance through iterations, to achieve efficient directed protein evolution. Extensive experiments have shown the effectiveness of our method in generating optimal mutants with minimal annotation effort, outperforming previous works even with fewer annotated mutants, making it budget-friendly for biological experiments.

Molecular representation learning contributes to multiple downstream tasks such as molecular property prediction and drug design. To properly represent molecules, graph contrastive learning is a promising paradigm as it utilizes self-supervision signals and has no requirements for human annotations. However, prior works fail to incorporate fundamental domain knowledge into graph semantics and thus ignore the correlations between atoms that have common attributes but are not directly connected by bonds. To address these issues, we construct a Chemical Element Knowledge Graph (KG) to summarize microscopic associations between elements and propose a novel Knowledge-enhanced Contrastive Learning (KCL) framework for molecular representation learning. KCL framework consists of three modules. The first module, knowledge-guided graph augmentation, augments the original molecular graph based on the Chemical Element KG. The second module, knowledge-aware graph representation, extracts molecular representations with a common graph encoder for the original molecular graph and a Knowledge-aware Message Passing Neural Network (KMPNN) to encode complex information in the augmented molecular graph. The final module is a contrastive objective, where we maximize agreement between these two views of molecular graphs. Extensive experiments demonstrated that KCL obtained superior performances against state-of-the-art baselines on eight molecular datasets. Visualization experiments properly interpret what KCL has learned from atoms and attributes in the augmented molecular graphs.

Electron beam inspection (EBI) with high resolution is a promising technique to improve the defect inspection on the surface of patterned wafer. However, high resolution usually means long inspection time, which results in the low throughput and limitation of EBI applied in practice. This study aims to optimize the inspection time of EBI by reducing the total number of inspection regions without loss of the accuracy. We first refine this defect inspection optimization problem as a partial congruent square cover problem. Then, we propose two novel mixed-integer linear programming models for this problem. To deal with the large-scale problems, an approximation algorithm is developed to obtain the high-quality solutions. This approximation algorithm efficiently utilizes the linear programming (LP) rounding technique and greedy strategy based on the proposed model. Compared with the existing algorithms in the literature, numerical results show the superiority of the proposed model and algorithm.Note to Practitioners—Defect inspection is a key process in wafer fabrication for identifying and inspecting the patterning defects generated during the complicated fabrication processes. Electron beam inspection (EBI) takes place of optical inspection gradually as the design rules keep shrinking and the circuits are more susceptible to nanoscale killer defects. Low throughput is the main drawback of EBI and the improvements on throughput have far-reaching significance on the high volume manufacturing of semiconductor products. This study aims to reduce the number of inspection regions to improve the total inspection time in the EBI process. Considering that the inspection time for each inspection region is constant, less inspection regions means less total inspection time. However, the positions of inspection regions are arbitrary across the continuous planar space, putting pressure on modeling and solving the problem. A preprocessing algorithm is designed to discover a limited number of candidate positions of inspection regions without loss of optimality, which greatly simplifies the problem. For dealing with large-scale instances, an approximation algorithm combining linear programming (LP)-rounding technique and greedy strategy is designed to get near optimal solutions. Since the number of inspection regions is one key factor determining the total inspection time, the proposed optimization methods have a potential to be applicable to enhance the efficiency of advanced EBI platforms, such as ASML HMI eP series.

This article investigates a hybrid flow shop scheduling problem that consists of a batch processor in the upstream and a discrete processor in the downstream. Limited waiting time between the batch processor and discrete processor is taken into consideration. Such a scheduling problem is commonly seen as bottlenecks in the production of precision parts, back-end process of semiconductor products, and glass and steel industries. A mixed-integer linear programming model is presented to minimize the makespan. Considering the complexity of this problem and the imperative requirement in real-time optimization, we first develop a constructive heuristic together with the worst case analysis by exploiting the key decision structure of the problem. Based on the decision structure, we then develop a learning-based scheduling approach via customized genetic programming to automatically generate effective heuristics for this problem. Lower bounds are also developed to provide a measurement for the performance of proposed algorithms. Numerical results show that our proposed algorithms outperform the existing metaheuristics and are capable of providing high-quality solutions using less computational time. Note to Practitioners-The production system consisting of a batch processor in the upstream and a discrete processor in the downstream is common in practice. The batch processor first handles a group of jobs simultaneously. Then, the jobs are released to a buffer to wait for the process on the discrete processor one by one. However, the waiting time of the jobs in the buffer is often required to be limited according to the production requirements. For example, after being heated in the heat-treatment oven, the aerospace precision parts have to be processed on the machining equipment in limited waiting time to improve the processability in subsequent manufacturing. The semiconductor chips have to be packed in limited waiting time after baking to avoid getting wet. The incongruous production modes between the batch processor and discrete processor, together with the limited waiting time constraint, make such operations always the bottleneck in manufacturing. Efficient heuristics, providing high-quality solutions with low time complexity, are much preferred in practice for most of the complicated scheduling problems, such as the scenarios described earlier. However, the designing process of an effective heuristic is tedious, and the heuristic is usually deeply customized for a certain production scenario. Genetic programming (GP) provides an inspiring approach to automatically generate sophisticated heuristics for complicated scheduling problems through evolutionary learning processes. By customizing a GP-based approach, the designing process of heuristics is automated, and some undetectable knowledge relations can be obtained to enhance the quality of heuristics. Such an approach facilitates to obtain more sophisticated schedules by analyzing valuable knowledge for smart manufacturing. The superiority of the heuristic learned by GP is shown in the computational experiment, and it has great potential to be applied to the practical scheduling.

We address the problem of full body human pose estimation in video. Most previous work consider body part, pose or trajectory of body part as basic unit to compose the pose sequence. In contrast, we consider tracklet of body part as the basic unit. Based on this medium granularity representation we develop a spatio-temporal graphical model to select an optimal tracklet for each part in each video segment. In our model, tracklet nodes of symmetric parts are coupled to one node to overcome the double counting problem. Through iterative spatial and temporal parsing, optimal solution is achieved in polynomial time. We apply our model on three publicly available datasets and show remarkable quantitative and qualitative improvements over the state-of-the-art approaches.

Foreground detection plays a fundamental role in video analysis. Frames with only background information are usually beneficial for many foreground detection algorithms, especially for regression-based methods where the background is recovered from a background basis matrix. However, many regression-based methods ignore the basis selection process or select bases by simple sampling, which may limit their performance. In this paper, a regression-based foreground detection method with a novel background basis selection process is proposed. The proposed basis selection method, which includes basis matrix construction and basis matrix update processes, aims to build an effective background basis matrix which helps to boost the performance of our foreground detection method. In our algorithm, the basis matrix construction process first builds the basis matrix locally with a multiple clustering evaluation process. With the locally constructed basis matrix, a modified linear regression-based foreground detection method is proposed for separating foreground and background globally. To further increase the representativeness and the adaptiveness of the background basis matrix, a basis matrix update algorithm is designed to incrementally replace the ineffective bases with new selected ones. Extensive experiments on challenging sequences demonstrate the effectiveness and the advantages of our method.

Our goal is to estimate contour flow (the contour pairs with consistent point correspondence) from inconsistent contours extracted independently in two video frames. We formulate the contour flow estimation locally as a motion segmentation problem where motion patterns grouped from optical flow field are exploited for local correspondence measurement. To solve local ambiguities, contour flow estimation is further formulated globally as a contour alignment problem. We propose a novel two-staged strategy to obtain global consistent point correspondence under various contour transitions such as splitting, merging and branching. The goal of the first stage is to obtain possible accurate contour-to-contour alignments, and the second stage aims to make a consistent fusion of many partial alignments. Such a strategy can properly balance the accuracy and the consistency, which enables a middle-level motion representation to be constructed by just concatenating frame-by-frame contour flow estimation. Experiments prove the effectiveness of our method.

Foreground detection with dynamic background is a challenging task in video surveillance analysis. When clean background bases are constructed, regression based foreground detection usually becomes more effective. In this paper, a novel basis selection method based on local neighborhood structure is proposed. The present method first constructs local neighborhood relationships among the basis candidates in a reconstruction manner. Then a multiple clustering strategy is designed to evaluate these basis candidates on local neighborhood structure. According to the evaluation score given by multiple clustering process, clean background bases (including dynamic background) are separated from candidates corrupted by foreground. By adding the proposed basis selection process to a modified linear regression framework, the foreground detection can be implemented in a more effective way. Experimental results on multiple videos show that the modified framework with basis selection is competitive with the state of the art.

This paper presents an overview and development of 2-D micromachined thermal wind sensors. It first compares different types of wind sensors, then gives a brief introduction to three basic measurement principles of the thermal wind sensors: hot-wire or hot-film, calorimetric, and time of flight. In order to have a good understanding of development of the 2-D micromachined thermal wind sensors, they are first categorized into thermoresistive, thermoelectric, and thermoelectronic wind sensors based on their operations. Then, each category is further divided into different subcategories, according to their measuring principles and sensor configurations. In Section IV, different materials, fabrication processes, packaging techniques, and control modes of the 2-D micromachined thermal wind sensors are analyzed. This paper also evaluates the core technologies on how to improve the sensitivity, measurement range, and power consumption of the thermal wind sensors.