Model merging is an efficient empowerment technique in the machine learning community that does not require the collection of raw training data and does not require expensive computation. As model merging becomes increasingly prevalent across various fields, it is crucial to understand the available model merging techniques comprehensively. However, there is a significant gap in the literature regarding a systematic and thorough review of these techniques. This survey provides a comprehensive overview of model merging methods and theories, their applications in various domains and settings, and future research directions. Specifically, we first propose a new taxonomic approach that exhaustively discusses existing model merging methods. Secondly, we discuss the application of model merging techniques in large language models, multimodal large language models, and more than ten machine learning subfields, including continual learning, multi-task learning, few-shot learning, and so on. Finally, we highlight the remaining challenges of model merging and discuss future research directions. A comprehensive list of papers about model merging is available at https://github.com/EnnengYang/Awesome-Model-Merging-Methods-Theories-Applications .

Recommender models are hard to evaluate, particularly under offline setting. In this article, we provide a comprehensive and critical analysis of the data leakage issue in recommender system offline evaluation. Data leakage is caused by not observing global timeline in evaluating recommenders e.g., train/test data split does not follow global timeline. As a result, a model learns from the user-item interactions that are not expected to be available at the prediction time. We first show the temporal dynamics of user-item interactions along global timeline, then explain why data leakage exists for collaborative filtering models. Through carefully designed experiments, we show that all models indeed recommend future items that are not available at the time point of a test instance, as the result of data leakage. The experiments are conducted with four widely used baseline models—BPR, NeuMF, SASRec, and LightGCN, on four popular offline datasets—MovieLens-25M, Yelp, Amazon-music, and Amazon-electronic, adopting leave-last-one-out data split. ¹ We further show that data leakage does impact models’ recommendation accuracy. Their relative performance orders thus become unpredictable with different amount of leaked future data in training. To evaluate recommendation systems in a realistic manner in offline setting, we propose a timeline scheme, which calls for a revisit of the recommendation model design.

Safeguarding drinking water is a major public health and environmental concern because it is essential to human life but may contain pollutants that can cause illness or harm the environment. Therefore, continuous research is necessary to improve water treatment methods and guarantee its quality. As part of this study, the effectiveness of coagulation–flocculation treatment using aluminum sulfate (Al₂(SO₄)₃) was evaluated on a very polluted site. Samplings were taken almost every day for a month from the polluted site, and the samples were characterized by several physicochemical properties, such as hydrogen potential (pH), electrical conductivity, turbidity, organic matter, ammonium (NH⁺₄), phosphate (PO₄³⁻), nitrate (NO₃⁻), nitrite (NO₂⁻), calcium (Ca²⁺), magnesium (Mg²⁺⁾, total hardness (TH), chloride (Cl⁻⁾, bicarbonate (HCO₃⁻), sulfate (SO₄²⁻), iron (Fe³⁺), manganese (Mn²⁺), aluminum (Al³⁺), potassium (K⁺), sodium (Na⁺), complete alkalimetric titration (TAC), and dry residue (DR). Then, these samples were treated with Al₂(SO₄)₃ using the jar test method, which is a common method to determine the optimal amount of coagulant to add to the water based on its physicochemical characteristics. A mathematical model had been previously created using the support vector machine method to predict the dose of coagulant according to the parameters of temperature, pH, TAC, conductivity, and turbidity. This Al₂(SO₄)₃ treatment step was repeated at the end of each month for a year, and a second characterization of the physicochemical parameters was carried out in order to compare them with those of the raw water. The results showed a very effective elimination of the various pollutions, with a very high rate, thus demonstrating the effectiveness of the Al₂(SO₄)₃. The physicochemical parameters measured after the treatment showed a significant reduction in the majority of the physicochemical parameters. These results demonstrated that the coagulation–flocculation treatment with Al₂(SO₄)₃ was very effective in eliminating the various pollutions present in the raw water. They also stress the importance of continued research in the field of water treatment to improve the quality of drinking water and protect public health and the environment.