Show Thoughts

Feldman dismisses de Kamps' and van der Velde's approaches to neural variable binding stating that they don't work for the general case "where new entities and relations can be dynamically added".