Meet Team Captricity: Angshuman “Angshu” Guha
September 23, 2014 by Angshuman Guha
Hi, I’m Angshu.
I recently joined Captricity as Chief Scientist. In past lives, I worked on handwriting recognition at Microsoft for over a decade and spent another decade or so at Google and Yandex Labs working on web search and machine learning.
So what does a Chief Scientist actually do? Good question! Here’s a bit more about what I’m up to and why I decided to join team Captricity.
At Captricity, we like to say we bridge people and machines.
For example, the process of extracting structured text from scanned images involves a hybrid of human workers as well as auto-magical computation -- the latter based on machine learning algorithms.
To be clear, this is not the old idea of AI that was once imagined to mimic and perhaps even exceed human intelligence. Rather, modern machine learning is more of a bottom-up process. Its purpose isn’t necessarily to understand human cognition, but to detect patterns from examples and “learn” from them -- so as to make interesting predictions for future examples. One obvious instance of machine learning we use is OCR, but we also employ intelligent automatic learning whenever possible. If we have many examples of a multiple-choice question in a form, for instance, we can learn a program that can automatically decipher marked answers on forms we haven’t seen before … cool, right? It’s a hard problem to solve, but it’s also very exciting and gratifying work!
Whenever possible, we leverage the latest machine learning practices here at Captricity, including support vector machines, artificial neural networks and so on, and we continue to be inspired by the spectacular successes achieved recently by deep learning methods in computer vision and natural language processing … it really is an awesome, exciting space!
If you can’t already tell, we believe the future of machine learning holds a ton of promise and I’m excited to be part of a talented team helping to pave the way forward!
Until next time,