Yuan Sun

Yuan Sun, Winton Nathan-Roberts, Tien Dung Pham et al.

In biomanufacturing, developing an accurate model to simulate the complex dynamics of bioprocesses is an important yet challenging task. This is partially due to the uncertainty associated with bioprocesses, high data acquisition cost, and lack of data availability to learn complex relations in bioprocesses. To deal with these challenges, we propose to use a statistical machine learning approach, multi-fidelity Gaussian process, for process modelling in biomanufacturing. Gaussian process regression is a well-established technique based on probability theory which can naturally consider uncertainty in a dataset via Gaussian noise, and multi-fidelity techniques can make use of multiple sources of information with different levels of fidelity, thus suitable for bioprocess modeling with small data. We apply the multi-fidelity Gaussian process to solve two significant problems in biomanufacturing, bioreactor scale-up and knowledge transfer across cell lines, and demonstrate its efficacy on real-world datasets.

Yuan Sun, Sisi Liu, Junjie Deng et al.

The pre-trained language model is trained on large-scale unlabeled text and can achieve state-of-the-art results in many different downstream tasks. However, the current pre-trained language model is mainly concentrated in the Chinese and English fields. For low resource language such as Tibetan, there is lack of a monolingual pre-trained model. To promote the development of Tibetan natural language processing tasks, this paper collects the large-scale training data from Tibetan websites and constructs a vocabulary that can cover 99.95$\%$ of the words in the corpus by using Sentencepiece. Then, we train the Tibetan monolingual pre-trained language model named TiBERT on the data and vocabulary. Finally, we apply TiBERT to the downstream tasks of text classification and question generation, and compare it with classic models and multilingual pre-trained models, the experimental results show that TiBERT can achieve the best performance. Our model is published in http://tibert.cmli-nlp.com/

Xiaofa Liu, Jianqin Yin, Yuan Sun et al.

In this paper, we develop an efficient multi-scale network to predict action classes in partial videos in an end-to-end manner. Unlike most existing methods with offline feature generation, our method directly takes frames as input and further models motion evolution on two different temporal scales.Therefore, we solve the complexity problems of the two stages of modeling and the problem of insufficient temporal and spatial information of a single scale. Our proposed End-to-End MultiScale Network (E2EMSNet) is composed of two scales which are named segment scale and observed global scale. The segment scale leverages temporal difference over consecutive frames for finer motion patterns by supplying 2D convolutions. For observed global scale, a Long Short-Term Memory (LSTM) is incorporated to capture motion features of observed frames. Our model provides a simple and efficient modeling framework with a small computational cost. Our E2EMSNet is evaluated on three challenging datasets: BIT, HMDB51, and UCF101. The extensive experiments demonstrate the effectiveness of our method for action prediction in videos.

Yonghao Dang, Jianqin Yin, Liyuan Liu et al.

Multi-person pose estimation (MPPE) presents a formidable yet crucial challenge in computer vision. Most existing methods predominantly concentrate on isolated interaction either between instances or joints, which is inadequate for scenarios demanding concurrent localization of both instances and joints. This paper introduces a novel CNN-based single-stage method, named Dual-path Hierarchical Relation Network (DHRNet), to extract instance-to-joint and joint-to-instance interactions concurrently. Specifically, we design a dual-path interaction modeling module (DIM) that strategically organizes cross-instance and cross-joint interaction modeling modules in two complementary orders, enriching interaction information by integrating merits from different correlation modeling branches. Notably, DHRNet excels in joint localization by leveraging information from other instances and joints. Extensive evaluations on challenging datasets, including COCO, CrowdPose, and OCHuman datasets, showcase DHRNet's state-of-the-art performance. The code will be released at https://github.com/YHDang/dhrnet-multi-pose-estimation.

Wenbin Gan, Yuan Sun

The ever growing abundance of learning traces in the online learning platforms promises unique insights into the learner knowledge assessment (LKA), a fundamental personalized-tutoring technique for enabling various further adaptive tutoring services in these platforms. Precise assessment of learner knowledge requires the fine-grained Q-matrix, which is generally designed by experts to map the items to skills in the domain. Due to the subjective tendency, some misspecifications may degrade the performance of LKA. Some efforts have been made to refine the small-scale Q-matrix, however, it is difficult to extend the scalability and apply these methods to the large-scale online learning context with numerous items and massive skills. Moreover, the existing LKA models employ flexible deep learning models that excel at this task, but the adequacy of LKA is still challenged by the representation capability of the models on the quite sparse item-skill graph and the learners' exercise data. To overcome these issues, in this paper we propose a prerequisite-driven Q-matrix refinement framework for learner knowledge assessment (PQRLKA) in online context. We infer the prerequisites from learners' response data and use it to refine the expert-defined Q-matrix, which enables the interpretability and the scalability to apply it to the large-scale online learning context. Based on the refined Q-matrix, we propose a Metapath2Vec enhanced convolutional representation method to obtain the comprehensive representations of the items with rich information, and feed them to the PQRLKA model to finally assess the learners' knowledge. Experiments conducted on three real-world datasets demonstrate the capability of our model to infer the prerequisites for Q-matrix refinement, and also its superiority for the LKA task.

Junjie Deng, Hanru Shi, Xinhe Yu et al.

Pre-trained language models are trained on large-scale unsupervised data, and they can fine-turn the model only on small-scale labeled datasets, and achieve good results. Multilingual pre-trained language models can be trained on multiple languages, and the model can understand multiple languages at the same time. At present, the search on pre-trained models mainly focuses on rich resources, while there is relatively little research on low-resource languages such as minority languages, and the public multilingual pre-trained language model can not work well for minority languages. Therefore, this paper constructs a multilingual pre-trained model named MiLMo that performs better on minority language tasks, including Mongolian, Tibetan, Uyghur, Kazakh and Korean. To solve the problem of scarcity of datasets on minority languages and verify the effectiveness of the MiLMo model, this paper constructs a minority multilingual text classification dataset named MiTC, and trains a word2vec model for each language. By comparing the word2vec model and the pre-trained model in the text classification task, this paper provides an optimal scheme for the downstream task research of minority languages. The final experimental results show that the performance of the pre-trained model is better than that of the word2vec model, and it has achieved the best results in minority multilingual text classification. The multilingual pre-trained model MiLMo, multilingual word2vec model and multilingual text classification dataset MiTC are published on http://milmo.cmli-nlp.com/.

Yuan Sun, Navid Salami Pargoo, Taqiya Ehsan et al.

Complex human activity recognition (CHAR) remains a pivotal challenge within ubiquitous computing, especially in the context of smart environments. Existing studies typically require meticulous labeling of both atomic and complex activities, a task that is labor-intensive and prone to errors due to the scarcity and inaccuracies of available datasets. Most prior research has focused on datasets that either precisely label atomic activities or, at minimum, their sequence approaches that are often impractical in real world settings.In response, we introduce VCHAR (Variance-Driven Complex Human Activity Recognition), a novel framework that treats the outputs of atomic activities as a distribution over specified intervals. Leveraging generative methodologies, VCHAR elucidates the reasoning behind complex activity classifications through video-based explanations, accessible to users without prior machine learning expertise. Our evaluation across three publicly available datasets demonstrates that VCHAR enhances the accuracy of complex activity recognition without necessitating precise temporal or sequential labeling of atomic activities. Furthermore, user studies confirm that VCHAR's explanations are more intelligible compared to existing methods, facilitating a broader understanding of complex activity recognition among non-experts.

Yuan Sun

For pre-training of MoE (Mixture-of-Experts) models, one of the main issues is unbalanced expert loads, which may cause routing collapse or increased computational overhead. Existing methods contain the Loss-Controlled method and the Loss-Free method, where both the unbalanced degrees at first several training steps are still high and decrease slowly. In this work, we propose BIP-Based Balancing, an expert load balancing algorithm based on binary integer programming (BIP). The algorithm maintains an additional vector q on each MoE layer that can help change the top-K order of s by solving a binary integer programming with very small time costs. We implement the algorithm on two MoE language models: 16-expert (0.3B) and 64-expert (1.1B). The experimental results show that on both models comparing with the Loss-Controlled method and the Loss-Free method, our algorithm trains models with the lowest perplexities, while saves at least 13% of pre-training time compared with the Loss-Controlled method. Within our current knowledge, this is the first routing algorithm that achieves maintaining load balance status on every expert in every MoE layer from the first step to the last step during the whole pre-training process, while the trained MoE models also perform well. The code material of this work is available at https://github.com/sunyuanLLM/bip_routing_algorithm.

Hongjie Xu, Yunzhuang Shen, Yuan Sun et al.

Column generation (CG) is a powerful technique for solving optimization problems that involve a large number of variables or columns. This technique begins by solving a smaller problem with a subset of columns and gradually generates additional columns as needed. However, the generation of columns often requires solving difficult subproblems repeatedly, which can be a bottleneck for CG. To address this challenge, we propose a novel method called machine learning enhanced ant colony optimization (MLACO), to efficiently generate multiple high-quality columns from a subproblem. Specifically, we train a ML model to predict the optimal solution of a subproblem, and then integrate this ML prediction into the probabilistic model of ACO to sample multiple high-quality columns. Our experimental results on the bin packing problem with conflicts show that the MLACO method significantly improves the performance of CG compared to several state-of-the-art methods. Furthermore, when our method is incorporated into a Branch-and-Price method, it leads to a significant reduction in solution time.

Yunzhuang Shen, Yuan Sun, Xiaodong Li et al.

Column generation (CG) is a well-established method for solving large-scale linear programs. It involves iteratively optimizing a subproblem containing a subset of columns and using its dual solution to generate new columns with negative reduced costs. This process continues until the dual values converge to the optimal dual solution to the original problem. A natural phenomenon in CG is the heavy oscillation of the dual values during iterations, which can lead to a substantial slowdown in the convergence rate. Stabilization techniques are devised to accelerate the convergence of dual values by using information beyond the state of the current subproblem. However, there remains a significant gap in obtaining more accurate dual values at an earlier stage. To further narrow this gap, this paper introduces a novel approach consisting of 1) a machine learning approach for accurate prediction of optimal dual solutions and 2) an adaptive stabilization technique that effectively capitalizes on accurate predictions. On the graph coloring problem, we show that our method achieves a significantly improved convergence rate compared to traditional methods.

Zecang Gu, Xiaoqi Sun, Yuan Sun et al.

Unsupervised clustering algorithm can effectively reduce the dimension of high-dimensional unlabeled data, thus reducing the time and space complexity of data processing. However, the traditional clustering algorithm needs to set the upper bound of the number of categories in advance, and the deep learning clustering algorithm will fall into the problem of local optimum. In order to solve these problems, a probabilistic spatial clustering algorithm based on the Self Discipline Learning(SDL) model is proposed. The algorithm is based on the Gaussian probability distribution of the probability space distance between vectors, and uses the probability scale and maximum probability value of the probability space distance as the distance measurement judgment, and then determines the category of each sample according to the distribution characteristics of the data set itself. The algorithm is tested in Laboratory for Intelligent and Safe Automobiles(LISA) traffic light data set, the accuracy rate is 99.03%, the recall rate is 91%, and the effect is achieved.

Yuan Sun, Sheng Wang, Yunzhuang Shen et al.

This paper introduces an enhanced meta-heuristic (ML-ACO) that combines machine learning (ML) and ant colony optimization (ACO) to solve combinatorial optimization problems. To illustrate the underlying mechanism of our ML-ACO algorithm, we start by describing a test problem, the orienteering problem. In this problem, the objective is to find a route that visits a subset of vertices in a graph within a time budget to maximize the collected score. In the first phase of our ML-ACO algorithm, an ML model is trained using a set of small problem instances where the optimal solution is known. Specifically, classification models are used to classify an edge as being part of the optimal route, or not, using problem-specific features and statistical measures. The trained model is then used to predict the probability that an edge in the graph of a test problem instance belongs to the corresponding optimal route. In the second phase, we incorporate the predicted probabilities into the ACO component of our algorithm, i.e., using the probability values as heuristic weights or to warm start the pheromone matrix. Here, the probability values bias sampling towards favoring those predicted high-quality edges when constructing feasible routes. We have tested multiple classification models including graph neural networks, logistic regression and support vector machines, and the experimental results show that our solution prediction approach consistently boosts the performance of ACO. Further, we empirically show that our ML model trained on small synthetic instances generalizes well to large synthetic and real-world instances. Our approach integrating ML with a meta-heuristic is generic and can be applied to a wide range of optimization problems.