2022 Fiscal Year Annual Research Report
Towards reaction-network-based reservoir computers as controllers for molecular robots
| Project/Area Number |
20K12061
|
|---|
| Research Institution | Ochanomizu University |
|---|
Principal Investigator |
オベル加藤 ナタナエル お茶の水女子大学, 基幹研究院, 講師 (10749659)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
GENOT Anthony 東京大学, 生産技術研究所, 国際研究員 (00761975)
|
|---|
| Project Period (FY) |
2020-04-01 – 2023-03-31
|
| Keywords | Molecular Robotics / Reservoir Computing / Thermosensing / Quality-Diversity / PEN DNA Toolbox / Surrogate models |
|---|
| Outline of Annual Research Achievements |
Following the objective mentioned at the beginning of the year, we implemented a microfluidic system providing a temperature range as input to our molecular reservoir. As a demonstration of the system, we implemented a "Spanish flag" corresponding to two types of behavior (A or B) distributed spatially in an A-B-A pattern. That naming is a reference to the "French flag" which corresponds to an A-B-C pattern. However, the A-B-A pattern, despite seemingly simpler, is more difficult to implement as it requires strong non-linearities in the system. An article summarizing our findings is in preparation.
On the theoretical side, we continued our exploration of the characteristics of molecular reservoirs based on the PEN DNA toolbox. We used our findings to develop a surrogate model for predicting the memory capacity and kernel rank (similar to computing power) of a given reservoir. While less accurate than a direct simulation, our surrogate model improves the evaluation speed of candidate solutions when designing a reservoir for a specific task. Our findings have been submitted as a research article.
Finally, we designed a novel molecular oscillator by extending our current molecular framework. The new system can be toggled between two different limit cycles, allowing the encoding of additional information into its behavior. Oscillators are among the best reservoir systems according to our current findings. As such, we expect this new system to further improve the capacities of molecular reservoir computing.
|
Research Products
(6 results)
