Difference between revisions of "Team:SEU/Human Practices"

Overview

Molecular computing technology, a new cross-disciplinary field of the information science, gradually attracts researchers’ attention due to its low computational complexity and high parallelism. Some researchers have designed molecular computation systems to achieve functions which are originally realized by traditional silicon-based systems (e.g., clock design and logic gates), which inspires us to think that maybe we can use molecules to realize artificial intelligence systems which is a very hot topic in recent years. With this goal, we first learned about artificial intelligence and then consulted specialist in Purple Mountain Laboratory (PML) for more insights. After the discussion with the specialist, we understood the limitation of the implementation of artificial intelligence systems. We also knew that molecular computing is possible to solve these problems.

Integrated Human Practices

Our visit in Purple Mountain Laboratory

At the beginning of our project, systems, we visited the Purple Mountain Laboratory (PML) in Nanjing to obtain better understanding of artificial intelligence. During our visit in PML, we discussed with specialist Dr. Tan about artificial intelligence.

Dr. Tan told us that most artificial intelligence systems are implemented with silicon-based chips. When facing large systems, the area complexity and power cost of such chips are high. Many researchers are trying to reduce the complexity and cost of the silicon-based artificial intelligence systems. She also mentioned that it was meaningful for us to use bio-materials to implement artificial intelligence systems since molecular computation has higher computation parallelism and lower energy cost compared to silicon-based chips.

During the discussion, we also learned that neural networks are basic parts in artificial intelligence systems. If we want to use molecules to realize artificial intelligence systems, we first need to build molecular computation models for artificial neurons, including all the related arithmetic operations. Besides, the training of neural networks should also be accomplished in molecular level. Meanwhile, Dr. Tan pointed out that our work may be efficient in the biological field. Our idea using DNA as the basis for construction may have good biological characteristics, and may be useful in intelligent recognition of biological information.

Our contribution to synthetic biology

In order to make the DNA-based neural networks available to other researchers, we developed a software tool for users, which can generate DNA reaction models and relevant DNA sequences. With this tool, users can better understand and improve our work.

Rome was not built in a day and every great advance is made up of many small achievements. In our work, we not only focus on case study, but also propose useful models and tools. We hope our project can provide convenience and help to other researchers.