Cluster 12: Machine Learning: Can We Teach a Computer to Think?
Week 3
Monday (7/24)
Students kicked off their Monday morning with a lecture led by Professor Neima about Processing Data in which students were taught how to clean out any unwanted and incomplete data out of a dataset and correct any possible mistakes with labels within said dataset. With the rest of their time before lunch, students were given a short hour to finish their Image Classification Projects and practice their presentations. Immediately after lunch, students presented their projects, ranging from sketching competitions to identifying schizophrenia based on the trees people drew. Congratulations to Angie and Aamir for winning faculty choice for their digital schizophrenia test, Jyoti and Ronit for their project Sketch Showdown for winning people’s choice, and to Angelina and Samuel for their project Draw off! for also winning people’s choice. Cluster 12 also got started with their new project based on Natural Language Processing, a part of machine learning that utilizes data to generate language and text, much like ChatGPT.
Tuesday (7/25)
Cluster 12 started their morning off with an interesting discovery lecture by Dr. Henrik I. Christensen, an esteemed professor of computer science and engineering here in UCSD, who discussed the inner workings of robotics and, specifically, the complex nature of self-automated cars. Professor Niema followed up the guest lecture with a presentation about evaluation metrics, equations used to gauge the accuracy of a model. Some evaluation metrics students were taught include binary classification in which various equations are used to identify a dataset’s sensitivity, specificity, precision, and more; and clustering classification in which students learned how to find a cluster’s homogeneity, completeness, and v-measure. After Professor Neima’s extensive lecture, students were given lab time to work on their Natural Language Processing projects.
Wednesday (7/26)
Cluster 12 began their day brainstorming topics for their final projects, a combination of every technique student’s have learned thus far which closes out their experience and education at COSMOS. After choosing partners and deciding on what they wanted to work on, students began mapping out their processes and proposing their ideas to be greenlit. Afterwards, students listened to a lecture by Professor Ideker on using machine learning to detect gene abnormalities that could lead to cancer detection. After lunch, class photos were taken in front of the UCSD bear and in the computer science building. All the remaining class time was used for working on their Natural Language Processing projects.
Thursday (7/27)
The start off their morning, students attended COSMOS’ annual Women in STEM Panel in which various women in the engineering industry discussed their personal experiences as a minority within a male-dominsted scene. They explored their struggles in being treated differently, for both the better and the worse, solely due to their gender instead of their qualifications and accomplishments. Then, students submitted their final project ideas to get approved and presented their Natural Language Processing projects. Programs ranged from text generators based on books and movies using Markov chains to algorithms that could tell whether text was positive or negative based on sentiment analysis. After lunch, Cluster 12 took their cluster photos by the UCSD bear and the computer science building. All the time after lunch was dedicated to working on final projects. Thank you to Teacher Fellow Shirley for brining in lumpia!
