Data Science Talks #1

Event Information

Share this event

Date and Time

Location

Location

Magnimind Academy

830 Stewart Dr, #182

Sunnyvale, CA 94085

View Map

Event description
We will describe two cases where we had to get creative to obtain training datasets

About this Event

In the real world, training datasets are not as readily available as Kaggle would have you believe. In certain situations, they are extremely costly to obtain, while in others they are impossible. In this talk, I will describe two cases where we had to get creative to obtain training datasets. I want an argumentative environment where we can discuss our ideas.

Record linkage is a classic problem in statistics and machine learning. There are two datasets, with several fields in common, but no unique identifiers as match keys. Due to a lack of ground truth training dataset, unsupervised learning techniques are usually employed. However, we embarked on an ambitious project to create a training dataset using unsupervised learning, and feed that to a machine learning model which then learned to make predictions of match or no match.

In the second project, we had to extract handwritten information from checks. The hard part of the problem is recognizing fields like dates, signatures etc on a check. The costly way would be to have human raters mark each field with coordinates. However, we decided to create a fake training dataset where we automatically generated date fields (populated with handwritten dates) and used that to train a classifier that recognized real date fields on checks.

Agenda:

6:20 pm - 6:30 pm Arrival and socializing

6:30 pm - 6:40 pm Opening words

6:40 pm - 7:50 pm Ahmad Emad, "Data Science Talk #1"

7:50 pm - 8:00 pm Q&A

About Ahmad Emad:

Ahmad Emad is currently a Data Scientist at Google, where he applies experiment design and machine learning techniques to make Google Assistant more useful and user-friendly. Prior to Google, Ahmad was a Data Scientist at Capital One building fraud detection models. Ahmad has a Masters in Statistics from The George Washington University and a Bachelors in Computer Science and Economics from Colgate University. In his free time, Ahmad enjoys reading, hiking and trying out new food places.

Date and Time

Location

Magnimind Academy

830 Stewart Dr, #182

Sunnyvale, CA 94085

View Map

Save This Event

Event Saved