文献类型： 会议论文
第一作者： Marco Valentino
作者： Marco Valentino 1 ; Jordan Meadows 2 ; Lan Zhang 2 ; Andre Freitas 3 ;
作者机构： 1.Idiap Research Institute, Switzerland
2.Department of Computer Science, University of Manchester, United Kingdom
3.Idiap Research Institute, Switzerland ##Department of Computer Science, University of Manchester, United Kingdom ##Cancer Biomarker Centre, CRUK Manchester Institute, United Kingdom
会议名称： [ "Conference of The North American Chapter of The Association for Computational Linguistics: Human Language Technologies" , "Conference of the North American Chapter of the Association for Computational Linguistics"]
主办单位：
页码： 1446-1458
摘要： This paper investigates the possibility of approximating multiple mathematical operations in latent space for expression derivation. To this end, we introduce different multi-operational representation paradigms, modelling mathematical operations as explicit geometric transformations. By leveraging a symbolic engine, we construct a large-scale dataset comprising 1.7M derivation steps stemming from 61K premises and 6 operators, analysing the properties of each paradigm when instantiated with state-of-the-art neural encoders. Specifically, we investigate how different encoding mechanisms can approximate expression manipulation in latent space, exploring the trade-off between learning different operators and specialising within single operations, as well as the ability to support multi-step derivations and out-of-distribution generalisation. Our empirical analysis reveals that the multi-operational paradigm is crucial for disentangling different operators, while discriminating the conclusions for a single operation is achievable in the original expression encoder. Moreover, we show that architectural choices can heavily affect the training dynamics, structural organisation, and generalisation of the latent space, resulting in significant variations across paradigms and classes of encoders.
分类号： tp311
- 相关文献
作者其他论文 更多>>
-
An LLM-based Knowledge Synthesis and Scientific Reasoning Framework for Biomedical Discovery
作者：Oskar Wysocki;Magdalena Wysocka;Danilo S. Carvalho;Alex Bogatu;Danilo Gusicuma;Maxime Delmas;Harriet Unsworth;Andre Freitas
关键词：
-
Learning Disentangled Semantic Spaces of Explanations via Invertible Neural Networks
作者：Yingji Zhang;Danilo S. Carvalho;Andre Freitas
关键词：
-
Graph-Induced Syntactic-Semantic Spaces in Transformer-Based Variational AutoEncoders
作者：Yingji Zhang;Marco Valentino;Danilo S. Carvalho;Ian Pratt-Hartmann;Andre Freitas
关键词：
-
Multi-Relational Hyperbolic Word Embeddings from Natural Language Definitions
作者：Marco Valentino;Danilo S. Carvalho;Andre Freitas
关键词：
-
Estimating the Causal Effects of Natural Logic Features in Transformer-Based NLI Models
作者：Julia Rozanova;Marco Valentino;Andre Freitas
关键词：
-
Reasoning with Natural Language Explanations
作者：Marco Valentino;Andre Freitas
关键词：
-
Inference to the Best Explanation in Large Language Models
作者：Dhairya Dalal;Marco Valentino;Andre Freitas;Paul Buitelaar
关键词：
