TindML

Project Title

TindML

Project Area of Specialization Artificial IntelligenceProject Summary

A Cloud-Based Autonomous Machine Learning Platform that will be based on auto-sklearn. It will extend the auto-sklearn with regression and time-series functionalities. A UI interface like BigML or Weka will be provided. It will have a web-based interface while using the storage and computing resources of AWS (Amazon Web Services). Research on model selection and parameter optimization will be carried out to simplify the process and optimize the solution.

Project Objectives

To provide a cloud-based autonomous machine learning platform using auto-sklearn and optimizing the model selection and parameter optimization techniques already being used (Meta-Learning and Bayesian Parameter Optimization).

Project Implementation Method

The project will be carried in different phases as below:

1. Literature Review (1 Month)
2. Problem Statement
3. Design
4. POC implementation
5. Testing
6. Full Implementation

The project will be build using open stack technologies.  PyCharm will be used as an IDE for backend development on the open-source Linux platform.

Benefits of the Project

The Novice (Non-Expert) Users of Data Science and Machine Learning, Skilled Professionals in Data Science & Machine Learning, Analytics Industry.

The problem (complexity of Machine Learning / Data Science) is a hindrance to the wide-scale adaptability of it in the general public as well as in the Analytic Industry. The Solution will foster its wide-scale adaptation.

There is an observed trend of Machine Learning and Data Science application for predictive and prescriptive analytics in global as well as local industry. Whether it be academia, banking, insurance, manufacturing, distribution or retail, the use of data science and machine learning is widening. The audience not only includes the Analytics Professional, but also the managers and other IT professionals as well as novices in data science and machine learning.

Technical Details of Final Deliverable

The Final Output will be:

1. A web-based Autonomous Machine Learning Platform (A web-based Software)
2. A Research Paper
3. User Manual

As our project is R&D based, we need access to high computing resources such as fast computers, SSDs and may need access to GPU.

For full implementation, we will need to host the framework on AWS. We may need access to the AWS server for 2-3 months which will have an associated cost. The cost of the AWS server depends on their compute, memory and storage requirement. These requirements are not clear at this time which server we will be using for test deployment.  We may need a compute-optimized instance that has an on-demand cost plan. 

Cost plan of different compute-optimized servers is given below:

Instance

VCPU

ECU

Memory

Storage

Cost

Instance

Final Deliverable of the Project HW/SW integrated systemCore Industry ITOther Industries Education , Finance Core Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Quality Education, Industry, Innovation and InfrastructureRequired Resources

Elapsed time in (days or weeks or month or quarter) since start of the project	Milestone	Deliverable
Month 1	Review Current Solutions	Literature Review
Month 2	Understanding Library Algorithms	Problem Statement
Month 3	Reverse Engineering Algorithms	Analyzing Problem Statement
Month 4	Combining Features	Planning Problem Solution
Month 5	Prototype Completion	Design
Month 6	POC Implementation	Development
Month 7	Testing	Integration
Month 8	Improvement	Evolution
Month 9	Implementation	Deployemeent