Seminar Details
2025-09-25 (13:00) : Learning from logical constraints with Lower- and Upper bound arithmetic circuits
At Shannon, Maxwell a.105
Organized by Computer Science and Engineering
Speaker :
Alexandre Dubray (ICTEAM)
Abstract :
In this work focuses on the field of ๐ง๐๐ฎ๐ซ๐จ-๐ฌ๐ฒ๐ฆ๐๐จ๐ฅ๐ข๐๐ฅ๐๐๐ซ๐ง๐ข๐ง๐ (NeSy), which aims to bridge the gap between deep learning methods (neural) and the logical knowledge available in certain domains (symbolic). It has been accepted to the Main track at IJCAI 2025, one of the worldโs premier conferences on Artificial Intelligence.
Standard deep learning struggles with logic-based reasoning. One solution is to encode known constraints as ๐๐ซ๐ข๐ญ๐ก๐ฆ๐๐ญ๐ข๐ ๐๐ข๐ซ๐๐ฎ๐ข๐ญ๐ฌ to enable a gradient-based guidance of the parameters being learned. But, for logical knowledge that is too complex to be fully encoded, existing methods use a single lower-bound approximate circuit, often compromising the quality of the computed gradients.
The authors introduce a ๐๐ฎ๐๐ฅ-๐๐จ๐ฎ๐ง๐ ๐๐ฉ๐ฉ๐ซ๐จ๐๐๐ก, ๐ฎ๐ฌ๐ข๐ง๐ ๐๐จ๐ญ๐ก ๐ ๐ฅ๐จ๐ฐ๐๐ซ- ๐๐ง๐ ๐๐ง ๐ฎ๐ฉ๐ฉ๐๐ซ-๐๐จ๐ฎ๐ง๐ ๐๐ข๐ซ๐๐ฎ๐ข๐ญ, to tightly control the error in the gradient approximation. This improves the robustness and trustworthiness of constraint-based learning.