Pareto-Based Multiobjective Machine Learning: An Overview and Case Studies
By Yaochu Jin, Bernhard Sendhoff. IEEE Transactions on Systems Man and Cybernetics Part C (Applications and Reviews), 2008
467 citations recorded by the source corpus.
Subfields: Artificial Intelligence, Computational Theory and Mathematics
View Paper: https://www.semanticscholar.org/paper/50feafd2cdafdfb1eead14388f6210f6b467eaa0
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
In Computational Theory and Mathematics
Citations in the corpus
- Modeling the nexus between carbon dioxide emissions and economic growth (2015, background): “A Pareto front is suitable for simultaneously balancing fitting accuracy and model complexity (Jin and Sendhoff, 2008).”
- An Evolutionary Algorithm for Large-Scale Sparse Multiobjective Optimization Problems (2020, methodology): “4) Neural Network Training Problem: Evolutionary algorithms have been deeply studied for training artificial neural network, such as optimizing topology [42], weights [43], and hyperparameters [20] of the network.”
- Metaheuristic design of feedforward neural networks: A review of two decades of research (2017, background): “The advantages of applying Pareto-based learning is thoroughly explained and compared with a single and scalerized objective in [295].”
- Efficient Large-Scale Multiobjective Optimization Based on a Competitive Swarm Optimizer (2020, background): “As suggested in [65], each solution is fine tuned by gradient descent after being generated for better approximation performance.”
- Global optimization issues in deep network regression: an overview (2018, background): “It can be viewed as a form of scalarization of the two conflicting objectives described by the bound (3) undergoing the learning procedure (see e.g. [ 83]).”
- Interpretable Rule Discovery Through Bilevel Optimization of Split-Rules of Nonlinear Decision Trees for Classification Problems (2020, methodology): “In [19], an ANN is used as a final classifier and a multi-objective optimization approach is employed to find simple and accurate classifier.”
- Learning to Abstain from Binary Prediction (2016, background): “Thm. 8 may be of independent interest as a complete minimax characterization of a bicriterion optimization problem; many learning problems can be written as multicriterion optimizations, but are typically scalarized ([CBL06, JS08]).”
- Multi-Objective Evolutionary Federated Learning (2018, background): “Direct binary encoding such as the one introduced in [32] needs a large connection matrix to represent the structure of a neural network, which is not scalable to neural networks containing multiple hidden layers and a large number of neurons.”
- Comprehensibility of Classification Trees–Survey Design (2019, background): “Although model complexity is related to comprehensibility [9], empirical user studies [1] reveal that comprehensibility measures based solely on model complexity are over-simplistic and produce incomprehensible models [5].”