A Life Cycle view of ML: Addressing Cardinal Issues
Mr Tarun Bonu1, Mrs Komathy Padmanabhan1, Mr Gnana Bharathy2
1Monash University, Clayton, Australia, 2ARDC, Melbourne, Australia
In order to explore key issues in AI/ML, we propose to stage a Workshop to start with a short presentation/ roundtable discussion on some of the most pressing issues in AI/ ML and then open the floor for a discussion. We aim to explore the key practices, challenges, limitations, and workarounds, relationships between various ideas grounded in practitioner experience.
A data science life cycle defines the stages or steps in a machine learning or data science project. We will use this life cycle view to classify key activities. The workshop will be split along these life cycle stages. For each stage, a mini-workshop will be conducted. Each workshop will have a presentation followed by a discussion relating to example use cases in that particular life cycle.
Note: While there are multiple life cycle depictions such as CRISP-DM, Team Life Cycle, Domino etc, in the following, we have adopted the common stages.
- Business Understanding and Framing the Problem
- Data Acquisition, Pre-Processing, and Exploration
- Feature Engineering, Model Development and Validation
- PoC, Customer Acceptance, PoC to Production
Deployment, Monitoring
- Pan-Cycle Issues: Explanation, AI Risk, Ethics, Participation, Value Creation and Measurement
Who should attend: AI/ML Practitioners. Researchers, Data scientists, Managers, Student Researchers
Skills or knowledge attendees should have: Some understanding of Machine learning will be useful
Do attendees need to install any special software to participate in the workshop: Participants will need to have access to Zoom with Audio and Visuals enabled
Pre workshop preparation: NA