2015 Fiscal Year Research-status Report
簡潔な・利用しやすい構造を有する学習ネットワークの構成と応用に関する研究
| Project/Area Number |
15K18095
|
|---|
| Research Institution | Waseda University |
|---|
Principal Investigator |
Parque Victor 早稲田大学, 理工学術院, 助教授 (50745221)
|
|---|
| Project Period (FY) |
2015-04-01 – 2019-03-31
|
| Keywords | networks / succint representation / learning / robotics / design |
|---|
| Outline of Annual Research Achievements |
The foundations for Genetic Counterfactual Programming (GCP) and a number of application benchmarks in computer-aided design and robotics were proposed: (1) The succinct, canonical and efficient representations of directed graphs was proposed, enabling graph encodings using single integers and network-based learning algorithms using number theory. (2) The benchmark for the application on aesthetic surfaces were established, wherein a novel proposed metric for aestheticism rendered smoother surfaces, significantly faster (124x improvement). (3) The benchmark for the application in parametric design problems under historic real-world data was established. This work enables the use of past data to evaluate novelty and performance in parametric conceptual designs, avoiding the use of inaccurate simulations or expensive real-world experiments, e.g. vehicle design problem. (4) The concepts of synchronous and concurrency were explored to develop a small-scale and low-cost sensor that detects the inclination of the arm accurately. This work makes possible the construction of affordable sensor interfaces for amputees in developing countries. (5) The foundations for representing networks with heterogeneous processing nodes and optimal / efficient communication topologies were established; in which an efficient algorithm was proposed to optimize ZigBee network topologies, enabling the construction of low-cost networks for multi-agent robots and the Internet of Things efficiently.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
The key foundations and benchmarks on how to represent and use Genetic Counterfactual Programming in design and robotics problems were established.
|
| Strategy for Future Research Activity |
In 2016, while gaining awareness and insights toward the deployment in real-world systems, research on the structure and the applications of Genetic Counterfactual Programming are to be conducted. Concretely, research on the following are to be conducted: (1) Structure and application of Genetic Counterfactual Programming in robotics: path planning, and (2) Structure and application of Genetic Counterfactual Programming in computer aided design: aesthetic curves.
|
| Causes of Carryover |
The reasons for incurring the amount to be used in next fiscal year are as follows: (1) Regarding Article Costs, savings are due to the fact of postponing the acquisition of a computer for numerical computations to the next fiscal year. (2) Regarding Travel Expenses, the cost savings are due to fact of attending near yet relevant conferences. Furthermore, the original tickets and receipts of travel expenses during 2015 are to be submitted in the next fiscal year. (3) Regarding miscellaneous costs, the journal articles corresponding to the extensions / developments of the ideas presented at conferences in the fiscal year 2015 are under submission / preparation. Also, original receipts of the registration fees of 2015 are to be submitted in the next fiscal year.
|
| Expenditure Plan for Carryover Budget |
It is expected that the amount to be used in the next fiscal year is split among the travel costs and registration fees of conferences attended during 2015, the costs for acquiring a computer suitable for sophisticated numerical computations and the costs for publishing at relevant journals venues.
|
