2018 Fiscal Year Annual Research Report
Study on Succint Learning Networks and its Applications
| Project/Area Number |
15K18095
|
|---|
| Research Institution | Waseda University |
|---|
Principal Investigator |
Parque Victor 早稲田大学, 理工学術院, 准教授(任期付) (50745221)
|
|---|
| Project Period (FY) |
2015-04-01 – 2019-03-31
|
| Keywords | learning networks / design / control / optimization / evolutionary computation / representations / combinatorial / pattern recognition |
|---|
| Outline of Annual Research Achievements |
In 2018, the algorithms and applications underlying Genetic Counterfactual Programming (GCP) in computer-aided design and robotics were developed: (1) the new algorithms for computing succinct networks in polygonal maps, enabling the enhanced scalability and efficiency in cluttered environments; (2) the new algorithms for representing general combination objects succinctly, realizing the utmost efficiency in time and space complexity; (3) the applications of learning networks with variable size in the data-driven design of vehicle layouts, enabling the generation of new vehicle models outperforming the frontiers of mileage consumption; (4) the applications of newly developed log-aesthetic polynomial curves in path planning, realizing the fast computation of smoother/safer curves for transportation/navigation; (5) the learning networks for (prosthetic) hand robotic interfaces and human activity recognition, achieving the highest accuracy in challenging testing environments; (6) the algorithms for coordinated synchrony realizing the seamless generation of flexible control rules in complex and changing environments; (7) the new learning algorithms that evolve hierarchical networks to generate decentralized/distributed/non-overlapping modules, whose application is potential to tackle the design of complex graphs/networks with best quasi-linear time complexity; (8) the new optimization algorithms for computing the succinct representation of large/distributed graphs and networks, achieving the encoding with utmost performance efficiency (in some cases, using 1 bit).
|
| Remarks |
The research on "On Minimal Trees in Polygonal Maps" was selected as finalist in the 15th Annual Humies Awards for Human-Competitive Results.
|
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
[Book] Computational Science2018
Author(s)
Yong Shi, Haohuan Fu, Yingjie Tian, Valeria V. Krzhizhanovskaya, Michael Harold Lees, Jack J. Dongarra, Peter M. A. Sloot
Total Pages
730
Publisher
Springer
ISBN
978-3-319-93697-0
