International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 11 Issue: 06 | Jun 2024
www.irjet.net
p-ISSN: 2395-0072
A Comparative Study of Automated Machine Learning Systems Against Manual Approaches S.B.S.S. Bharadwaj1, P. Thrilochan Reddy2, Dr. Diana moses3 1Student, Dept of Artificial Intelligence and Data Science, Methodist college of engineering and technology,
Telangana, India 2Student, Dept of Artificial Intelligence and Data Science, Methodist college of engineering and technology,
Telangana, India 3Professor, Dept of Computer Science and Engineering, Methodist college of engineering and technology,
Telangana, India ------------------------------------------------------------------------***----------------------------------------------------------------------
Abstract- Machine learning (ML) is a computers ability
model for a given data set with hyperparameter optimization to a traditional machine learning approach. This system aims to streamline the ML pipeline from data processing to model evaluation using the latest technology in algorithm selection, hyperparameter optimization and performance evaluation. Through comprehensive evaluation, we aim to compare the performance of our Auto-ML system with manual methods and evaluate its performance in terms of predictive accuracy, computational efficiency, and ease of use.
to help computer systems recognize patterns from data and draw conclusions from it. It allows machines to learn from data and make better decisions using various algorithms without being explicitly programmed. In the current era of artificial intelligence and machine learning, choosing the best model and adjusting its hyperparameters for the available variety of data is a very chaotic process. Here comes automated machine learning systems (Auto-ML), which provide a dynamic way to streamline this process and help overcome this challenge. By utilizing a variety of assessment measures including accuracy, RMSE, and MAE values, our study provides a clear comparison of the Auto-ML technique to the conventional manual approach. This research gives us a strong knowledge of the computational effectiveness and scalability of the models for a given data.
The importance of this research lies in its potential to revolutionize the landscape of ML model development and deployment. By automating the labour-intensive and error-prone aspects of model selection and hyperparameter tuning, AutoML can democratize ML knowledge, allowing industry professionals of diverse experience to harness the power of advanced predictive analytics. Additionally, by systematically comparing Auto-ML with manual approaches, we try to provide insight into the trade-offs involved and identify scenarios where each approach excels.
Key Words- Hyperparameter Tuning, Artificial Intelligence, Auto-ML, Accuracy, computational efficiency.
1.INTRODUCTION In recent years, the proliferation of machine learning (ML) applications in various fields has emphasized the importance of effectively selecting of appropriate models and fine-tuning hyperparameters for optimal performance.
Here, we are going to use different machine learning models like RandomForest classifier, logistic regression, Support Vector Machines and K-Nearest Neighbour classifier for manual approaches and .For Auto-ML approaches, we will use the H2O Auto-ML model and the Auto-WEKA tool. We are going to use the most popular IRIS dataset for both the approaches. We will compare and evaluate using various parameters like model accuracy, RMSE and MAE values.
The traditional approach to this task often involves manual research, where data scientists repeatedly try different algorithms and hyperparameter settings, a process that consumes both time and resources. However, the emergence of automatic machine learning (Auto-ML) systems offers a promising alternative by automating these labour-intensive tasks, which can democratize ML adoption and accelerate innovation.
This research project contributes to the ongoing Auto-ML debate by introducing a new approach to automatic model selection and hyperparameter tuning. By empirically evaluating the performance of our Auto-ML system compared to manual approaches, we aim to
The goal of this research is to compare an Auto-ML system that can automatically select the best-fit ML
© 2024, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 421