Intro -- Preface -- Organization -- Keynote Talks -- Unsupervised Model Selection in Outlier Detection: The Elephant in the Room -- Coloring Social Relationships -- 35 Years of 'Scientific Discovery: Computational Explorations of the Creative Processes' - From the Early Days to the State of the Art -- Contents -- Regression and Limited Data -- Model Optimization in Imbalanced Regression -- 1 Introduction -- 2 Related Work -- 3 Imbalanced Regression -- 3.1 Relevance Function -- 3.2 Squared Error Relevance Area (SERA) -- 4 Optimization Loss Function for Imbalanced Regression
5 Experimental Study -- 5.1 Experimental Setup -- 5.2 Results on Model Optimization -- 5.3 Results in Out-of-Sample -- 6 Conclusions -- A SERA numerical approximation -- B Tables of Results -- References -- Discovery of Differential Equations Using Probabilistic Grammars -- 1 Introduction -- 2 Related Work -- 3 Methods -- 3.1 Algebraic Equations and Numeric Differentiation -- 3.2 Differential Equations and Direct Simulation -- 3.3 Parallel Computation -- 4 Experimental Evaluation -- 4.1 Experimental Setup -- 4.2 Results -- 5 Conclusion -- References
Hyperparameter Importance of Quantum Neural Networks Across Small Datasets -- 1 Introduction -- 2 Background -- 2.1 Functional ANOVA -- 2.2 Supervised Learning with Parameterized Quantum Circuits -- 3 Methods -- 3.1 Hyperparameters and Configuration Space -- 3.2 Assessing Hyperparameter Importance -- 3.3 Verifying Hyperparameter Importance -- 4 Dataset and Inclusion Criteria -- 5 Results -- 5.1 Performance Distributions per Dataset -- 5.2 Surrogate Verification -- 5.3 Marginal Contributions -- 5.4 Random Search Verification -- 6 Conclusion -- References
ImitAL: Learned Active Learning Strategy on Synthetic Data -- 1 Introduction -- 2 Simulating AL on Synthetic Training Data -- 3 Training a Neural Network by Imitation Learning -- 3.1 Imitation Learning -- 3.2 Neural Network Input and Output Encoding -- 3.3 Pre-selection -- 4 Evaluation -- 4.1 Experiment Details -- 4.2 Comparison with Other Active Learning Strategies -- 5 Conclusion -- References -- Incremental/Continual Learning -- Predicting Potential Real-Time Donations in YouTube Live Streaming Services via Continuous-Time Dynamic Graph -- 1 Introduction -- 2 Related Work
2.1 Online Live Streaming Service -- 2.2 Dynamic Graph Learning -- 3 Methodology -- 3.1 Dataset -- 3.2 Dynamic Graph Generation -- 3.3 Temporal Graph Neural Network -- 3.4 Strategies for Data Imbalance -- 4 Experiments -- 4.1 Dataset Description -- 4.2 Experiment Setup -- 4.3 Baselines -- 4.4 Evaluation -- 4.5 Case Study -- 5 Conclusion -- References -- Semi-supervised Change Point Detection Using Active Learning -- 1 Introduction -- 2 AL-CPD -- 2.1 Algorithm Outline -- 2.2 Selecting Candidate Change Points -- 2.3 Finding New Candidate Change Points -- 3 Experiments -- 3.1 Datasets
Summary
This book constitutes the proceedings of the 25th International Conference on Discovery Science, DS 2022, which took place virtually during October 10-12, 2022. The 27 full papers and 12 short papers presented in this volume were carefully reviewed and selected from 59 submissions.