This past week I continued to research my topic and the differences between the neural network methods that could be used to model aqueous solubility. This was my first research assessment regarding the details of neural network modeling, which should be useful to me once I start designing my own models. I was surprised to learn about the strong predictive power that even algorithms that are not machine learning-based display, suggesting that cheminformatics may increasingly replace traditional methods of calculating chemical properties. In my future research, I want to examine the performance of the same types of neural networks for other applications. For example, would more layered neural networks be better suited for complex problems like drug discovery?
I realized that I need to learn more about applying the algorithms I researched by translating them into code while also expanding my knowledge of Python. At the same time, I need to spend more time this week researching the chemistry side of my topic, especially the modeling tools commonly used in cheminformatics. Luckily, I was able to schedule an interview with a professional chemist, which I was very happy about. I’m excited to learn more about my topic from him and other professionals so I can hopefully find a mentor sooner rather than later.
Comments