Exploring Explainability in Machine Learning

Dear all,

Here is the schedule for your upcoming one-on-one meetings. Please make a note of your assigned date and time. Ensure that you are punctual and prepared for the meeting.

Monday 25 Nov. 2024 in Building E1 3, Room 523

• 4:00 PM - 4:20 PM: Shashank Priyadarshi

Tuesday 26 Nov. 2024 in Building E1 3, Room 523

• 9:00 AM - 9:20 AM: Maurice Vincon

Friday 29 Nov. 2024  in Building E1 3, Room 528

• 2:00 PM - 2:20 PM: Nhi Pham
• 2:25 PM - 2:45 PM: Juan José Valenzuela González
• 2:50 PM - 3:10 PM: Tulika Nayak
• 3:15 PM - 3:30 PM: Shreyansh Tripathi
• 3:35 PM - 3:55 PM: Mayur Deshmukh

If you have any questions or need to reschedule, please contact me as soon as possible.

Thank you!

Hanwei

Dear all,

I am pleased to announce that office hours are now accessible on the timetable for your convenience. The first scheduled office hour will be on Wednesday at 15:00 in Building E1 3, Room 523. Additional dates and times will be added soon, so please check the timetable regularly for updates.

Looking forward to meeting with you during these times to assist with any questions or guidance you may need.

Best regards,

Hanwei

Hey all,

Thank you all for taking the time to complete the topic bidding process! Your thoughtful selections reflect a strong interest in advancing our understanding of model explainability and interpretability, and we’re excited to see the unique perspectives you’ll bring to these advanced topics.

Here are your confirmed assignments:

• Juan José Valenzuela González– LIME: Model-agnostic Local Explanations for Complex Models
• Tulika Nayak – SHAP: Feature Importance Using Shapley Values from Game Theory
• Shashank Priyadarshi – LRP: Layer-wise Relevance Propagation
• Nhi Pham – Counterfactual Explanations: “What-if” Scenarios for Actionable Insights
• Mayur Deshmukh – Anchors: Rule-based, High-Precision Explanations for Stability
• Shreyansh Tripathi – Graph Neural Networks Explainability: Explaining GNNs for Structured Data
• Maurice Vincon – Concept Activation Vectors (TCAV): High-Level, Human-Friendly Explanations

Now that topics are finalized, please start by reviewing the assigned paper, available through the links provided in the slides from our kick-off course. As you investigate the paper and related concepts in explainable AI, gather relevant information to structure your research. Then, begin developing your assignments, focusing on clarity, depth, and originality.

Resources & Support
Slides are available in the course materials section on the website. If you have any questions or need further support, please don’t hesitate to reach out!

Important:
Please register for the "examination" in the LSF by November 22 to ensure your participation is confirmed.

Thank you once again for your participation in the topic selection process. I am eager to see the insights and creativity each of you will bring to these exciting areas of machine learning explainability.

Best regards,

Hanwei