Instead of writing a weekly update on my project, I would like to take a moment to reflect on what I have been able to achieve in Quarter 3. I will also discuss the plans I have for Q4 as well as my final product.
Constant Shift of Focus
Throughout the Q3, I have frequently experienced shifts of focus on my project. Originally, my plan was to further my understanding of machine learning and possibly train a model on my own. I was able to watch a few lessons on Coursera. However, as I had become increasingly concerned about the sustainability of Polaris, I started shifting my focus to Polaris about 2 weeks into the quarter. After finalizing all the proposed Polaris features, I started doing some consulting for projects of current CS 1A students. One of the projects uses the new Polaris API to display timetable on the library’s monitor. Another is set to be released during the next software release event. It has been quite rewarding for me to be able to touch on so many aspects of computer science. If I could redo what I got to do in the first quarter, I would have done it in a more organized fashion.
Problem with Coursera
While I really enjoy Andrew Ng’s Intro to Machine Learning course on Coursera, I do not appreciate the pace at which it is taught. Coursera imposes a period of learning time on courses. Once I went past that period, my progress (quizzes, availability of videos, etc.) was reset so that I had to restart the entire course. I am currently using Google’s machine learning crash course as it gives me more flexibility. It also uses TensorFlow, a machine learning software library I have always wanted to explore. I will certainly revisit the coursera course in the near future.
Plan for Q4
Essentially, my plan for Q4 would be the same as originally planned with the following exceptions:
- Learn machine learning with Google instead of Coursera
- Learn TensorFlow as opposed to other frameworks
- Consulting for Vega (unreleased product)
- Plan for software release event
- Focus on social entrepreneurship
The Final Product
While enjoying the beautiful views in Yunnan, China, I also had the chance to brainstorm a couple of ideas for my final product. Ideally, the product of my choice can relate to social entrepreneurship, which will also count toward my completion of the Social Entrepreneurship Certificate at Westtown. Here, I would like to unveil a potential candidate.
I stumbled upon a video made by Google about using used cellphone to prevent illegal logging activities in the Amazon forest. Google collaborate with local rangers to train a machine learning model that identifies the sound of chainsaw and deployed to the cellphones with TensorFlow. The cellphones are installed in the trees as collection points and are powered by solar panels.
The idea of recycling used cellphones alone is socially beneficial. When equipped with software that turns them into powerful data collection points, these cellphones can help address challenges that cannot be solved otherwise. My final product is likely to be a solution that addresses a social challenge using a network of recycled cellphones.
In next week’s blog, I will write about my experience with installing TensorFlow. Thanks for reading and see you next week!
Cell phones in Amazon trees alert rangers to illegal logging, record wildlife. Digital Trends, http://www.digitaltrends.com/outdoors/solar-powered-cell-phones-saving-amazon-rainforests/. Accessed 2 Apr. 2018.