Article
Computer Science, Artificial Intelligence
I Bakurov, M. Castelli, F. Fontanella, A. Scotto di Freca, L. Vanneschi
Summary: Geometric semantic genetic programming (GSGP) is a variant of genetic programming that transforms the landscape of any supervised regression problem into a unimodal error surface. In a previous study, a novel variant of GSGP was proposed for binary classification problems, which showed promising performance by using a logistic-based activation function to constrain the output value. This paper presents the results of 18 test problems and compares them with other well-known classification schemes, demonstrating the effectiveness of the proposed approach.
SWARM AND EVOLUTIONARY COMPUTATION
(2022)
Article
Computer Science, Artificial Intelligence
Noman Javed, Fernand Gobet, Peter Lane
Summary: Genetic programming suffers from the problem of excessive growth in individuals' sizes, which reduces its ability to explore complex search spaces efficiently. This paper focuses on reviewing the literature from an explainability perspective and how simplification can make GP models more explainable by reducing their sizes. Researchers have proposed various simplification techniques, and this paper organizes the literature to identify their strengths and weaknesses, as well as emerging trends and areas for future exploration.
DATA MINING AND KNOWLEDGE DISCOVERY
(2022)
Article
Computer Science, Artificial Intelligence
Ying Bi, Bing Xue, Mengjie Zhang
Summary: A new Genetic Programming based approach is proposed in this article for automatically learning informative features for different image classification tasks. The approach uses a flexible program structure to evolve solutions of variable depths, extracting various numbers and types of features from images. The results demonstrate that the new approach achieves better classification performance than most benchmark methods.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2021)
Article
Optics
Jie Liu, Da Song, Guohua Geng, Yu Tian, Mengna Yang, Yangyang Liu, Mingquan Zhou, Kang Li, Xin Cao
Summary: In this paper, a task-driven and learnable down-sampling method named TGPS is proposed to address the issue of excessive redundant data in the dense point clouds of Terracotta Warriors obtained by a 3D scanner. The method utilizes a point-based Transformer unit to embed features and a mapping function to extract input point features for dynamic representation of global features. The contribution of each point to the global feature is estimated using the inner product between the global feature and each point feature, and high-similarity point features are retained based on descending contribution values. A Dynamic Graph Attention Edge Convolution (DGA EConv) is proposed for local feature aggregation, combined with graph convolution operation. Networks for point cloud classification and reconstruction are presented as downstream tasks. Experimental results demonstrate that the method achieves downsampling guided by global features, with TGPS-DGA-Net achieving the best accuracy on both real-world Terracotta Warrior fragments and public datasets.
Article
Computer Science, Artificial Intelligence
Qinglan Fan, Ying Bi, Bing Xue, Mengjie Zhang
Summary: This article proposes a new image classification approach using genetic programming (GP) with a new program structure that allows for flexible feature reuse. The approach can automatically learn various informative image features and select a suitable classification algorithm based on the learned features. Experimental results show that the approach achieves better performance than many state-of-the-art methods.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2023)
Article
Computer Science, Artificial Intelligence
Tapas Si, Pericles Miranda, Joao Victor Galdino, Andre Nascimento
Summary: This paper investigates the application of grammar-based automatic programming in medical data classification, which has implicit capabilities of automatic feature selection and extraction. Experimental results show that the GAP approach is able to produce suitable classifiers for given problems and outperforms classical classifiers.
ARTIFICIAL INTELLIGENCE REVIEW
(2021)
Article
Computer Science, Artificial Intelligence
Ying Bi, Bing Xue, Mengjie Zhang
Summary: This article introduces a multitask genetic programming approach for image feature learning, with a new knowledge sharing mechanism allowing automatic learning of shared content across tasks to enhance learning performance. Utilizing a new individual representation, each task is addressed using common and task-specific trees.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2022)
Article
Geosciences, Multidisciplinary
Pece V. Gorsevski
Summary: This research utilized genetic programming to predict landslide susceptibility spatially, achieving high accuracy through symbolic classification, digital elevation model, and topographic prediction attributes.
Article
Environmental Sciences
Miao Lu, Ying Bi, Bing Xue, Qiong Hu, Mengjie Zhang, Yanbing Wei, Peng Yang, Wenbin Wu
Summary: Information on crop spatial distribution is crucial for agricultural monitoring and food security. This study proposes a novel Genetic Programming (GP) approach to learn high-level features from time series images for crop classification. Experimental results demonstrate that the GP features can improve classification accuracy and are more robust and stable compared to traditional features and deep learning methods.
Article
Computer Science, Artificial Intelligence
Mohammad Beheshti Roui, Mariam Zomorodi, Masoomeh Sarvelayati, Moloud Abdar, Hamid Noori, Pawel Plawiak, Ryszard Tadeusiewicz, Xujuan Zhou, Abbas Khosravi, Saeid Nahavandi, U. Rajendra Acharya
Summary: This paper introduces a novel approach for generating classification rules based on evolutionary computation, with custom crossover and mutation operators for GPU execution, and leveraging parallelism to enhance fitness function performance. Experimental results demonstrate high accuracy and speedup for HCV, Poker, and COVID-19 datasets.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Wenbin Pei, Bing Xue, Lin Shang, Mengjie Zhang
Summary: This paper introduces a novel GP method for developing cost-sensitive classifiers by automatically learning cost matrices instead of requiring them from domain experts, demonstrating superior performance compared to existing GP methods in most cases.
APPLIED SOFT COMPUTING
(2021)
Article
Computer Science, Artificial Intelligence
Su Nguyen, Dhananjay Thiruvady, Mengjie Zhang, Kay Chen Tan
Summary: The article introduces a new genetic programming approach to optimize the search mechanism in constraint programming by evolving efficient variable selectors. The results demonstrate that evolved variable selectors can significantly reduce the computational effort of the search solver and increase the likelihood of finding optimal solutions.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2021)
Article
Computer Science, Artificial Intelligence
Wenlong Fu, Bing Xue, Xiaoying Gao, Mengjie Zhang
Summary: Transfer learning using Genetic Programming (GP) in document classification tasks successfully applies GP programs evolved from the source domain (SD) directly to classify documents in the target domain (TD), showing improved accuracy and performance.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Medical Informatics
Atharva Phatak, David W. Savage, Robert Ohle, Jonathan Smith, Vijay Mago
Summary: This study develops a deep learning-based text simplification approach for converting complex medical text into a simpler version while maintaining text quality. The proposed method outperforms previous baselines in terms of Flesch-Kincaid scores and achieves comparable performance in other metrics. The developed approach increases the readability of complex medical paragraphs, making biomedical research more accessible to a wider audience.
JMIR MEDICAL INFORMATICS
(2022)
Article
Environmental Sciences
Caiyun Wen, Miao Lu, Ying Bi, Shengnan Zhang, Bing Xue, Mengjie Zhang, Qingbo Zhou, Wenbin Wu
Summary: This paper proposes a new object-based Genetic Programming (GP) approach to extract cropland fields. The approach combines genetic programming with multiresolution segmentation technique to extract spectral, shape, and texture features of the fields and automatically evolves the optimal classifier. The results show that the proposed approach achieves high accuracy in areas with different landscape complexities and outperforms commonly used classifiers.