top of page
Engineering Service
A reflective log of engineering service activities
Talk on Machine Learning
Date: 22 Feb 2018
Time: 6:00 - 8:00 PM
Organizer: SCU ACM - Santa Clara University Association for Computing Machinery
Host: Computer History Museum NextGen Advisory Board
Topic: The Many Facets of Machine Learning: How ML works & Where It's Showing Up
The speakers driving the talk worked at Lyft, Deepscale, and Kyndi. The talk covered an introduction to convolutional neural networks, an overview of Natural Language Processing, and Machine Learning applications such as self-driving cars.
The biggest takeaway from this talk was when Paras Jain, a research engineer at DeepScale, mentioned one of the biggest problems in the Machine Learning application domain is dealing with the tradeoff between performance and accuracy. He had mentioned that better data is needed for better metrics which helps create better models. As a result, the focus of his research and efforts centered around accuracy, speed, and scale.
As a software developer, the solutions I have focused on so far take into account accuracy. This talk shed light on what constitutes an effective, efficient solution. The performance of a solution was something I have yet to explore and learn about. Next year I will be taking a class on algorithms which will cover Big-O Notation. I plan to understand how solutions can be improved in terms of performance and explore how can they be scaled as well. Eventually, I hope to build high-performance, accurate, deep scaled solutions for the problems I work on.
Workshop on Machine Learning
Date: 24 Feb 2018
Time: 11 AM - 1:30 PM
Organizer: SCU ACM - Santa Clara University Association for Computing Machinery
Topic: TensorFlow workshop
At this workshop, I got hands-on experience in a domain I had never explored before: Machine Learning. After learning the basics of Machine Learning at the talk on Thursday, the workshop on Saturday taught me how to build my own neural network to classify images of handwritten images.
The neural network I built had an accuracy of 96.5%. So, I got a taste of building a high-performance, accurate solution. Upon inquiring, I understood that the accuracy of the network could be increased by increasing the data set which in this case meant running the solution multiple times since the network could learn from itself and hence, would perform better the next time.
While I am currently focused on Web Development, I wish to explore Machine Learning eventually. Instead of viewing the different developing areas of Computer Science as separate entities, I realized at the end of the day, every area wants to create a high-performing, accurate, fast solution and so, one has the freedom to explore different areas at once. For instance, web development and machine learning. Web Applications which resort to conventional data mining methods could use machine learning instead.
This workshop made me reflect upon the intersection between web development and machine learning. This intersection is an opportunity I am excited to explore.
bottom of page