From: 2023-02-17 11:15 to 12:00
Place: Seminar Room KC 3N27 and Zoom
Contact: johan [dot] eker [at] control [dot] lth [dot] se

Jonas Lilja is defending his Master's Thesis at the Dept. of Automatic Control in Lund.

When: 17/2 -11:15-12:00
Where: Seminar Room KC 3N27 and Zoom: https://lu-se.zoom.us/j/62701218431
What: Master’s thesis presention
Title: Evaluating machine learning models for text classification - A comparative study of Amazon Comprehend & Amazon SageMaker
Speaker: Jonas Lilja
Advisor: Johan Eker, Dept. of Automatic Control, LTH
Examiner: Karl-Erik Årzén, Dept. of Automatic Control, LTH

Abstract: 
Together with Sigma Technology Cloud, I have conducted my master's thesis revolving around the use of cloud-based services on AWS for processing text data. More precisely, my thesis has been focused on automatically estimating the sentiment/attitudes for product and service reviews written in Swedish by applying different machine learning techniques.

The thesis evaluates Amazon's built-in solution (Amazon Comprehend) and compares it to NLP models built using Amazon's platform Amazon SageMaker. To build the models, different approaches and proven methods were tested to find the model that gives the best results in terms of accuracy and cost-effectiveness.
 
Finally, examining what benefits can be gained from constructing such a solution compared to the ready-made service. The contribution of the thesis thus becomes an exploration of systematic approaches for evaluating cloud deployments and benchmarking the solution's performances. With the hope that this contribution will enable organizations to make informed decisions about cloud utilization and adoption.