Assignment 7
Reading Response and Markov Chain
Reflections:
Both the interview with Emily Martinez, Nikita Huggins, and Ayodamola Okunseinde discussed questions about ML algorithms, datasets, and honor lineage. ML algorithms are essentially determined by the input dataset. To intentionally optimize the possible output, we, especially those of us who are not capable of working with the darkbox of ML algorithms yet, should focus on formatting and optimizing the input dataset. At the same time, it’s also where the filter should be placed. The morality and honor lineage have always been the controversial topic about ML datasets. Ideally, designing an algorithm that is capable of filtering immoral/unethical data would be the better approach. But practically, we often need to manually check the input datasets. However, within the current mainstream AI models, there is still a strong West-centric bias. It’s not necessarily intentional, but the phenomena are left to be resolved. To solve the problem at its root, there must be a more diverse AI developer community.
Markov Chain practices
(I'm still working on it, will update my progress within today)
Comments
Post a Comment