Bo-Cheng Lin
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Bo-Cheng Lin has a total of 3 co-authors. Here are the top co-authors by number of shared publications:
- Yi Mei
Centre for Data Science and Artificial Intelligence & School of Engineering and Computer Science, Victoria University of Wellington, New Zealand
2 shared publications Mengjie Zhang
Centre for Data Science and Artificial Intelligence & the School of Engineering and Computer Science Univer- sity of Wellington 6140 Wellington Victoria New Zealand
1 shared publication- Mengjie Zhang
Centre for Data Science and Artificial Intelligence & School of Engineering and Computer Science, Victoria University of Wellington, New Zealand
1 shared publication
Latest Publications
- Open Access
HEATACO: Heatmap-Guided Ant Colony Decoding for Large-Scale Travelling Salesman Problems
Heatmap-based non-autoregressive solvers for large-scale Travelling Salesman Problems output dense edge-probability scores, yet final performance largely hinges on the decoder that must satisfy degree-2 constraints and form a single Hamiltonian tour. Greedy commitment can cascade into irreparable mistakes at large $N$, whereas MCTS-guided local search is accurate but compute-heavy and highly engineered. We instead treat the heatmap as a soft edge prior and cast decoding as probabilistic tour construction under feasibility constraints, where the key is to correct local mis-rankings via inexpensive global coordination. Based on this view, we introduce HeatACO, a plug-and-play Max-Min Ant System decoder whose transition policy is softly biased by the heatmap while pheromone updates provide lightweight, instance-specific feedback to resolve global conflicts; optional 2-opt/3-opt post-processing further improves tour quality. On TSP500/1K/10K, using heatmaps produced by four pretrained predictors, HeatACO+2opt achieves gaps down to 0.11%/0.23%/1.15% with seconds-to-minutes CPU decoding for fixed heatmaps, offering a better quality--time trade-off than greedy decoding and published MCTS-based decoders. Finally, we find the gains track heatmap reliability: under distribution shift, miscalibration and confidence collapse bound decoding improvements, suggesting heatmap generalisation is a primary lever for further progress.Open AccessHEATACO: Heatmap-Guided Ant Colony Decoding for Large-Scale Travelling Salesman Problems
Heatmap-based non-autoregressive solvers for large-scale Travelling Salesman Problems output dense edge-probability scores, yet final performance largely hinges on the decoder that must satisfy degree-2 constraints and form a single Hamiltonian tour. Greedy commitment can cascade into irreparable mistakes at large $N$, whereas MCTS-guided local search is accurate but compute-heavy and highly engineered. We instead treat the heatmap as a soft edge prior and cast decoding as probabilistic tour construction under feasibility constraints, where the key is to correct local mis-rankings via inexpensive global coordination. Based on this view, we introduce HeatACO, a plug-and-play Max-Min Ant System decoder whose transition policy is softly biased by the heatmap while pheromone updates provide lightweight, instance-specific feedback to resolve global conflicts; optional 2-opt/3-opt post-processing further improves tour quality. On TSP500/1K/10K, using heatmaps produced by four pretrained predictors, HeatACO+2opt achieves gaps down to 0.11%/0.23%/1.15% with seconds-to-minutes CPU decoding for fixed heatmaps, offering a better quality--time trade-off than greedy decoding and published MCTS-based decoders. Finally, we find the gains track heatmap reliability: under distribution shift, miscalibration and confidence collapse bound decoding improvements, suggesting heatmap generalisation is a primary lever for further progress. - Open Access
- 28 January 2025
- Published by Springer Nature in Memetic Computing
- Vol. 17 (1) , 1-22
Automated design of state transition rules in ant colony optimization by genetic programming: a comprehensive investigation
The automated design of Ant Colony Optimization (ACO) algorithms has become increasingly significant, particularly in addressing complex combinatorial optimization problems. Although existing methods have achieved some success, they still face limitations, particularly the high dependency on expert knowledge, pre-solved data, and challenges in interpretability. Genetic Programming (GP), as a proven technology, has shown potential in optimizing the automated design state transition rules of ACO. However, existing research on GP-ACO is insufficient, particularly in terms of experimental validation and systematic evaluation. To address these issues, this study conducts comprehensive experiments to explore several key questions: the generality of GP-ACO on homogeneously distributed maps, the impact of different ACO variants on the learning capabilities of GP-ACO, the effect of 2-opt local search on the learning capabilities of GP-ACO, the enhancement of learning capabilities through the addition of more global information, and the interpretability of GP-ACO. The findings indicate that GP-ACO exhibits robust generality; variations among ACO variants have minimal impact on learning performance; 2-opt local search can somewhat diminish the performance of GP-ACO in the Max–Min Ant System; additional global information can significantly enhance the learning capabilities of GP-ACO; and GP-ACO has good interpretability.
- 28 January 2025
- Published by Springer Nature in Memetic Computing
- Vol. 17 (1) , 1-22
Open AccessAutomated design of state transition rules in ant colony optimization by genetic programming: a comprehensive investigation
The automated design of Ant Colony Optimization (ACO) algorithms has become increasingly significant, particularly in addressing complex combinatorial optimization problems. Although existing methods have achieved some success, they still face limitations, particularly the high dependency on expert knowledge, pre-solved data, and challenges in interpretability. Genetic Programming (GP), as a proven technology, has shown potential in optimizing the automated design state transition rules of ACO. However, existing research on GP-ACO is insufficient, particularly in terms of experimental validation and systematic evaluation. To address these issues, this study conducts comprehensive experiments to explore several key questions: the generality of GP-ACO on homogeneously distributed maps, the impact of different ACO variants on the learning capabilities of GP-ACO, the effect of 2-opt local search on the learning capabilities of GP-ACO, the enhancement of learning capabilities through the addition of more global information, and the interpretability of GP-ACO. The findings indicate that GP-ACO exhibits robust generality; variations among ACO variants have minimal impact on learning performance; 2-opt local search can somewhat diminish the performance of GP-ACO in the Max–Min Ant System; additional global information can significantly enhance the learning capabilities of GP-ACO; and GP-ACO has good interpretability.