Generative Pre-Trained Transformer for Design Concept Generation: An Exploration
نویسندگان
چکیده
Abstract Novel concepts are essential for design innovation and can be generated with the aid of data stimuli computers. However, current generative algorithms focus on diagrammatic or spatial that either too abstract to understand detailed early phase exploration. This paper explores uses pre-trained transformers (GPT) natural language concept generation. Our experiments involve use GPT-2 GPT-3 different creative reasonings in tasks. Both show reasonably good performance verbal
منابع مشابه
Concept Generation for Persuasive Design
Designing ‘persuasive’ products and services for social benefit often involves adopting and adapting principles and patterns from other disciplines and contexts where behaviour change is a goal. This poster briefly reports on a series of controlled trials of an idea generation toolkit which aims to make this transposition of patterns easier, with designers and students applying the toolkit to f...
متن کاملDesign for Survivability: Concept Generation and Evaluation in Dynamic Tradespace Exploration
Multi-Attribute Tradespace Exploration (MATE) for Survivability is introduced as a system analysis methodology to improve the generation and evaluation of survivable alternatives during conceptual design. MATE for Survivability applies decision theory to the parametric modeling of thousands of design alternatives across representative distributions of disturbance environments. To improve the ge...
متن کاملLeadership for Innovation in Healthcare: An Exploration
Although leadership has been studied extensively, most research has focused on the political and military spheres. More recent work has also examined the role of leadership in sectors such as manufacturing and technology, both areas where it is essential to encourage and nurture innovation. Yet, in the health sector, where innovation is now high on the policy agenda in many countries, there is ...
متن کاملLanguage Generation with Recurrent Generative Adversarial Networks without Pre-training
Generative Adversarial Networks (GANs) have shown great promise recently in image generation. Training GANs for text generation has proven to be more difficult, because of the non-differentiable nature of generating text with recurrent neural networks. Consequently, past work has either resorted to pre-training with maximumlikelihood or used convolutional networks for generation. In this work, ...
متن کاملDesign Exploration of Hybrid CMOS-OxRAM Deep Generative Architectures
Deep Learning and its applications have gained tremendous interest recently in both academia and industry. Restricted Boltzmann Machines (RBMs) offer a key methodology to implement deep learning paradigms. This paper presents a novel approach for realizing hybrid CMOS-OxRAM based deep generative models (DGM). In our proposed hybrid DGM architectures, HfOx based (filamentary-type switching) OxRA...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Proceedings of the Design Society
سال: 2022
ISSN: ['2732-527X']
DOI: https://doi.org/10.1017/pds.2022.185