A carregar...
Publicação
EasyML: An AutoML System using Meta Learning and Particle Swarm Optimization
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 1.68 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
In recent years machine learning has made great strides in many application areas and an ever-growing
number of disciplines rely on it. However, machine learning modelling process involves trying many
machine learning algorithms with different parameter configurations which is considered insufficient,
tedious, and time-consuming. The challenge has brought about the need for off-the-shelf solutions
that allow a dataset to choose its best modelling pipeline including data preprocessing, model selection
and hyperparameter optimization without or with very little human intervention in the process.
Despite the availability of numerous AutoML systems that can automate the machine learning
modeling process, there is still a need for a solution that can achieve the same results using a
significantly smaller space, while improving efficiency. This thesis proposes an AutoML system named
EasyML that uses meta-learning for model selection and particle swarm optimization for
hyperparameter optimization. The research objectives include conducting a comprehensive literature
review on State-of-the-Art techniques and existing AutoML systems, design, and development of
EasyML, evaluating the system's performance on benchmark datasets, comparing its efficiency to
other AutoML systems, and identifying its limitations and suggesting future research directions. The
research methodology combines Design Science Research and CRISP-DM. EasyML outperforms existing
solutions like SmartML and Auto-WEKA on all benchmark datasets. EasyML has the potential to
contribute to the development of more efficient and effective AutoML systems, thereby meeting the
increasing demand for data scientists with strong knowledge of various machine learning algorithms
and techniques.
Descrição
Dissertation presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Data Science
Palavras-chave
Automated Machine Learning Meta Learning Particle Swarm Optimization Hyperparameter Optimization OpenML