Embedded Files
All content is written, photographed, and edited by our amazing COSMOS students!
Week 3
Monday - Day 15
After a fun weekend at the San Diego Zoo, we returned back to our classes. We started out our day watching some Ethics Videos. After, we reviewed the Neural Networks we learned last Friday and learned Convolutional Neural Networks. Finishing that, we went into Natural Language Processing, specifically, Tokenization. To end off our lecture portion, we had an interesting guest lecture by Julian McAuley, about Recommender Systems.
After Lunch, we returned to our lab portion of the day. We started off with our Census Data Logistic Regression Presentations. Finishing those, we spent the rest of the day on our next project, about utilizing Neural Networks to classify images.
During ResLife, many of us headed to Gliderport for nightly programs. Others stayed back and played Just Dance, painted, or built their resume.
Tuesday - Day 16
We started off our Tuesday with watching the rest of everyone's Ethics Videos. After that, we brainstormed for our final project proposals. Near the end of the lecture period, we went over some sample abstracts, reading some research papers, and writing the abstracts for them, following a process given to us by Mrs. Miranda.
We spent the majority of our lab session on our Neural Networks Project. It was pretty difficult, and there were a lot of errors everywhere. Thankfully, we had our professors and TAs to help us overcome these rampant errors.
Wednesday - Day 17
Starting off early at 9 am, students dived into a long, 3-hour lecture with Professor Soosairaj (with breaks in between, of course). We learned more about NLP and reviewed Tokenization, Language Processing Pipelines, Stemming + Lemmetazization, Name Entity Recognition, and lastly, we were introduced to make our own spam-filtering models using Bags of Words and Naive Bayers.
Upon request, students got a WHOLE extra day to work on their MLP + CNN projects! Students got time to continue exploring the hyperparameters of their CNN models, implementing things like batch normalization, drop features, and data augmentation. Some students even went the extra mile and searched up YouTube tutorials to build complex models like ResNet 34, 50, and 152. Lastly, students prepared their presentations, ready to present the next day.
Thursday- Day 18
The day started with a "Women in STEM" panel, where students learned about the paths some engineers took to get to where they were today. They learned of several cases of sexism, and how the engineers overcame/dealt with them. They cemented the importance of making positive connections, seeking helpful mentors, and maintaining hope while exploring different professions in their lives. After, the students got more time to finalize their project proposal ideas and see some example poster board presentations.
Lab time started with presentations of the Neural Networks project. Everyone did very well, and some students even reached 96.6% accuracy! Following a short break, students started on their NLP Spam-filtering project.
Page updated
Report abuse