Week 1 went quickly! We met a lot of people and had an opportunity to learn about machine learning and the PATHS-UP project. We want to capture our experiences each week and have a time capsule of the things we have learned. To assist in arranging your thoughts, please reflect on the following questions:
- What were some of the things you did to prepare for the Young Scholars Program?
- Tell us some things you learned from the faculty/graduate student talks that you were able to sit in on.
- What are your first thoughts on the research projects that you have been assigned and your mentors?
- What are you most looking forward to for week two?
The first week at the PATHS-UP program with Rice was characterized by giving us a strong foundation to get started on machine learning. In order to prepare for the internship, I made sure to review python so I can be ready to work with ML algorithms, along with making sure that I had enough space on my laptop to download additional software. With our daily faculty/graduate student meetings, I learned the basics of being able to use GitHub to work on projects, learned the differences between supervised and unsupervised ML, how to work with scikit-learn, the causes of diabetes in underserved communities, and the fundamentals of ML regression. I am very interested to be working with supervised ML, as I hope to use it to work on computer vision for robotics. I look forward to continuing to get started on our supervised ML projects with our given data sets, along with getting to further learn about ML as a whole for week 2.
To prepare for the Young Scholars Program, I completed the readings of the resources we were sent prior to the program which went over machine learning terminology as well as using a breast cancer dataset. I also tried to brush up on Python a little since I did not have much experience with the programming language coming into the program. One of the most memorable moments from week 1 has been Dr. Roman’s presentation as she discussed her work as the co-lead of PATH-UP’s Thrust 3, or wearables. An astonishing fact that I learned fromthis presentation was that 40% if African American adults are more likely to become obese and for African American children, this percentage skyrocketed up to 73%, which seems unbelievable! This, accompanied by the other detailed information that Dr. Roman presented, made me better understand the significance of building health technologies that would directly account for and better serve marginalized communities. I also learned that in existing fitness/health wearables, the device error was higher for males, those with greater BMI and darker skin tone, and when walking. This further reinforced how important it is to build health technologies using data that well-represents the diversity of people instead of relying on skewed or biased data. Although I do not have extensive thoughts about our research projects since we have not fully dived into it yet, I will say that I felt a spark of thrill at the challenge while also feeling intimidated by the introduction of our unsupervised research project. However, on day 1, Ankit and Bhargav were reassuring and patient enough to explain the basics of machine learning and ease us into our research projects. My specific mentor, Bhargav, generously gave us a crash course on Python and patiently answered all of our questions, which eased some of the initial nervousness I had. Overall, my first week in this program was eye-opening and enlightening, and I am looking forward to developing our research project in more detail during week 2 since we have learned the foundations of machine learning in week 1!
Before the PATHS UP program, I had a feeling of almost nervous excitement. I was looking forward to the program and how to apply aspects of computer science to actual research. I read up on some of the articles our mentors sent before the program started and grasped what we would do in the upcoming weeks. I also downloaded and became familiar with Pycharm to use my time during the program effectively and not learn the editor during my time in the program. We had graduate meetings where they presented and briefed us on their research throughout our days. I found the diabetes presentation very interesting as my dad currently deals with it and how it’s genuinely a matter of diet and keeping up a healthy and active lifestyle. It was also interesting how socioeconomic backgrounds affect people’s susceptibility to diabetes due to not being able to afford a healthy diet and how some minorities are disproportionally affected by diabetes. As we went back to our workflow, we were presented with our project, and I think it’s incredible that we’re working with actual data and real problems. It gives me a taste of what actual research looks like. I believe mentors are excellent guides with our worm. They’re always there to help and are very understanding of situations and problems we may encounter. I never really see our mentors in a lousy mood. They’re always ready to teach and be as helpful as possible. Im looking forward to getting to the main course meal this week as we’ve been preparing everything to start working and implementing our data and algorithms to put them to work.
To prepare for the paths up young scholars program, I completed all the assigned reading regarding the basics of machine learning and I watched countless lessons on coding in python. I was definitely a little nervous going into this week. I had minimal experience with python and only a broad understanding of machine learning before starting this program. However, after completing week one, I am happy to report that I no longer feel unsure of my machine learning capabilities but rather feel empowered and excited to learn more. This week I have learned more than I could have ever imagined. Namely, Dr.Straach taught us that machine learning is responsible for many programs that we use in our daily lives. For example, she taught us that the facial recognition system on my iPhone’s photo album app uses a machine learning algorithm, specifically convolutional neural networks, to identify what faces appear in each of my photos. Furthermore, machine learning algorithms are responsible for the nifty Netflix recommendations that use data on the shows I have watched to determine other shows I may be interested in. Dr.Straach also explained that machine learning is best applied in fields where human expertise is limited. She explained that there are two main subsets to machine learning: supervised and unsupervised learning. Supervised learning uses labeled data sets and are best used to classify data or predict outcomes. Unsupervised learning attempts to decipher clusters or associations in data sets that are not labeled. Dr.Straach also introduced us to deep learning which is a subset of machine learning that attempts to simulate how the brain functions by using artificial neural networks. Deep learning makes use of many layers of information processing in order to train a computer using less human assumptions. Furthermore, Dr.Straach taught us about perceptron, Adaline, and logistic regression models which are highly valuable parametric models used in supervised learning machine learning projects.
In addition to the incredible knowledge I gained from Dr.Straach, I learned about the specific functions and code required to run a machine learning program from my mentors Bhargav and Ankit. Specifically, Bhargav taught us about the importance of analyzing, trimming, and understanding your data before starting to train the data. He advised us to use matplotlib to visualize our data and analyze our data for missing data.
He also introduced us to K-means clustering, an unsupervised machine learning algorithm, which allows a machine to find clusters in a large set of data. Bhargav has been incredibly helpful and understanding this week. I am very grateful to be working with his guidance.
One insightful lesson I learned from the graduate/professor talks was that type two diabetes rates are often correlated with certain races and income groups. Underprivileged individuals often do not have access to healthcare and healthy food which makes them more likely to suffer with diabetes. In addition, a lack of understanding and trust in the healthcare system makes it harder to treat underprivileged patients. I was amazed to see how machine learning can be applied to improve access to healthcare in underserved communities.
When I was first introduced to the project my group is working on, an unsupervised machine learning project on the breast cancer data set, I was definitely apprehensive. I was overwhelmed by the size of the data set and the different cluster/association algorithms in unsupervised learning. However, after closely analyzing the data set and having Bhargav explain the code required to complete a K-means clustering model I feel more comfortable with this project.
Finally, I look forward to learning about the other forms of clustering algorithms such as agglomerative clustering or affinity propagation. I also look forward to growing more comfortable with machine learning so I can work independently with ML techniques to support underprivileged community members in the future.
The week definitely did fly by. Every single moment was so interesting, and each day felt like the beginning of a new day of exciting discoveries and exploring a world I’d never seen before. When preparing for the PATHS-UP program, I went over the previous coding knowledge I had, spoke with friends who have experience with machine learning, went over the reading assigned, and researched the basics of machine learning. I reviewed the reading multiple times making sure to mark words and concepts I was unsure about and note them later on alongside their definitions. One of my favorite parts of the internship has been the faculty and graduate student talks. So far the research we have sat in on has been diabetes related, in specific making diabetes care more accessible for underserved populations. One of the biggest things I realized was how different the rate and effect of diabetes is on different groups of people, and how much more likely underserved populations are to develop diabetes, specifically type two diabetes. Another important thing I learned from the faculty talks was the immense role of technology such as smart watches in aiding these groups of people in successfully managing their diabetes. Another interesting aspect of the internship has been the research we are currently working on. About a couple months ago, my aunt was diagnosed with breast cancer that has since been removed. In her case, she was diagnosed very early, so treatment was efficient, and she was able to remove the cancer and is nearly recovered now. Ever since then, the importance of diagnosing cancer as early as possible has been extremely important to me. When I found out we would be working with a cancer data set to determine whether a tumor was malignant or benign, I was very interested. To be able to have the opportunity to make breast cancer diagnosis much more efficient is so important to me due to breast cancer being one of the most common cancers women are diagnosed with around the world. Another reaction I had to the research project was surprise. I had not realized quite how large a machine learning dataset was until just this week. Luckily, with the help of Bhargav Ghanekar, the mentor for the unsupervised learning group, we have been able to understand how to go about the creation of our machine learning program. Bhargav showed us many examples using the iris dataset to allow us to clearly visualize how we might use similar techniques for the breast cancer dataset. As we move into actually beginning the majority of the code for our programs this second week, I can easily state that is what I am most excited for. To be able to code the entire program, watch it run, and eventually watch it perform accurately is definitely something that I am ready to experience in this second week of the PATHS-UP internship.