Hareem Qazi


Master of Science in Computer Science


Department of Computer Science

Faculty / School

Faculty of Computer Sciences (FCS)

Date of Submission



Dr. Sajjad Haider, Professor, Department of Computer Science

Document type

MSCS Survey Report


Designing an effective end-to-end Machine Learning pipeline can prove to be a daunting task even for highly trained Data Scientists having a sound command over a specific domain. The computationally intensive task of finding a perfect combination of hyperparameters and an ML algorithm for a given data set led to the development of the Automated Machine Learning (AutoML) technique.

AutoML aims to empower domain experts to build ML pipelines, with minimal assistance from a Data Scientist, to understand a machine learning process, complex statistical concept or system architecture needs. AutoML models apply various meta-learning and hyperparameter tuning approaches by utilizing different tools to produce optimum results.

This research survey aims to highlight the developments in the Automated Machine Learning domain. The research is further narrowed down into AutoML models proposed for supervised and unsupervised learning problems.

Moreover, the survey summarizes and evaluates the features and functionalities of widely popular AutoML tools. Acknowledging AutoML's soaring demand, the paper identifies a few essential areas that need to be addressed by the research community and discusses future improvements to enhance the performance of AutoML models.

The full text of this document is only accessible to authorized users.