Redirigiendo al acceso original de articulo en 23 segundos...
Inicio  /  Applied Sciences  /  Vol: 14 Par: 7 (2024)  /  Artículo
ARTÍCULO
TITULO

CTGGAN: Controllable Text Generation with Generative Adversarial Network

Zhe Yang    
Yi Huang    
Yaqin Chen    
Xiaoting Wu    
Junlan Feng and Chao Deng    

Resumen

Controllable Text Generation (CTG) aims to modify the output of a Language Model (LM) to meet specific constraints. For example, in a customer service conversation, responses from the agent should ideally be soothing and address the user?s dissatisfaction or complaints. This imposes significant demands on controlling language model output. However, demerits exist among traditional methods. Promoting and fine-tuning language models exhibit the ?hallucination? phenomenon and cannot guarantee complete adherence to constraints. Conditional language models (CLM), which map control codes into LM representations or latent space, require training the modified language models from scratch and a high amount of customized dataset is demanded. Decoding-time methods employ Bayesian Rules to modify the output of the LM or model constraints as a combination of energy functions and update the output along the low-energy direction. Both methods are confronted with the efficiency sampling problem. Moreover, there are no methods that consider the relation between constraints weights and the contexts, as is essential in actual applications such as customer service scenarios. To alleviate the problems mentioned above, we propose Controllable Text Generation with Generative Adversarial Networks (CTGGAN), which utilizes a language model with logits bias as the Generator to produce constrained text and employs the Discriminator with learnable constraint weight combinations to score and update the generation. We evaluate the method in the text completion task and Chinese customer service dialogues scenario, and our method shows superior performance in metrics such as PPL and Dist-3. In addition, CTGGAN also exhibits efficient decoding compared to other methods.

 Artículos similares

       
 
Yulong Bai, Guolian Li, Tianxiu Lu, Yadong Wu, Weihan Zhang and Yidan Feng    
Most existing road network matching algorithms are designed based on previous rules and do not fully utilize the potential of big data and historical tracks. To solve this problem, we introduce a new road network matching algorithm based on deep learning... ver más
Revista: Applied Sciences

 
Shashidhar Rudregowda, Sudarshan Patil Kulkarni, Gururaj H L, Vinayakumar Ravi and Moez Krichen    
Visual speech recognition (VSR) is a method of reading speech by noticing the lip actions of the narrators. Visual speech significantly depends on the visual features derived from the image sequences. Visual speech recognition is a stimulating process th... ver más
Revista: Acoustics

 
Michael Tetteh, Allan de Lima, Jack McEllin, Aidan Murphy, Douglas Mota Dias and Conor Ryan    
Grammatical Evolution is a Genetic Programming variant which evolves problems in any arbitrary language that is BNF compliant. Since its inception, Grammatical Evolution has been used to solve real-world problems in different domains such as bio-informat... ver más
Revista: Algorithms

 
Ridwan Ilyas, Masayu Leylia Khodra, Rinaldi Munir, Rila Mandala and Dwi Hendratmo Widyantoro    
The paraphrase generator for citation sentences is used to produce several sentence alternatives to avoid plagiarism. Furthermore, the generation results need to pay attention to semantic similarity and lexical divergence standards. This study proposed t... ver más
Revista: Informatics

 
Anton Sysoev    
The construction of a mathematical model of a complicated system is often associated with the evaluation of inputs? (arguments, factors) influence on the output (response), the identification of important relationships between the variables used, and red... ver más
Revista: Computation