| Title | : | The Case for Circulatities in Case-Based Reasoning |
| Speaker | : | Parsodkar Adwait Pramod (IITM) |
| Details | : | Wed, 2 Aug, 2023 4:00 PM @ MR - I (SSB 233) |
| Abstract: | : | We aim at highlighting the importance of 'circularity' in diverse Case-Based Reasoning (CBR) tasks. We use the term 'circularity' to refer to settings where a property A can be determined only when a property B is known; however, determining property B, in turn, requires the knowledge of property A. Our central thesis is that a methodical examination of circularity can reveal gaps in maintenance and retrieval approaches used in CBR and pave the way for more effective algorithms. We further posit that novel approaches proposed for resolving circularity in a particular task can potentially provide insights for addressing circularity in a seemingly unrelated task. One instance of circularity is concerned with the estimation of reliability of cases in the case base, where our approach, RelCBR, that is founded upon a circular definition for the estimation of reliability is empirically shown to be more effective over the non-circular baseline approaches. Though the work till date has been restricted in scope to CBR, it holds promise in offering foundational insights in other areas of AI and Machine Learning as well. |
