Generative Diversity

Generative diversity focuses on enhancing the variety and representativeness of data generated by artificial intelligence models, aiming to improve model performance and mitigate biases. Current research emphasizes strategies like prompt engineering, variational autoencoders (VAEs), and conditional GANs (CGANs) to achieve this, often incorporating techniques to measure and optimize for both intra-class and inter-class diversity within generated datasets. This work is significant because improved generative diversity leads to more robust and fair AI systems across various applications, from instance segmentation and medical image analysis to text-to-image generation and design optimization.

Papers

May 16, 2024

DiverGen: Improving Instance Segmentation by Learning Wider Data Distribution with More Diverse Generative Data
Chengxiang Fan, Muzhi Zhu, Hao Chen, Yang Liu, Weijia Wu, Huaqi Zhang, Chunhua Shen
Generative Model Data Augmentation Instance Segmentation Data Distribution Generative Diversity

April 25, 2024

DE-CGAN: Boosting rTMS Treatment Prediction with Diversity Enhancing Conditional Generative Adversarial Networks
Matthew Squires, Xiaohui Tao, Soman Elangovan, Raj Gururajan, Haoran Xie, Xujuan Zhou, Yuefeng Li, U Rajendra Acharya
Conditional Generative Adversarial Network Generative Diversity Transcranial Magnetic Stimulation Functional Network Connectivity

January 30, 2024

Pick-and-Draw: Training-free Semantic Guidance for Text-to-Image Personalization
Henglei Lv, Jiayu Xiao, Liang Li, Qingming Huang
Text to Image Generative Question Pick and Place Semantic Guidance Personalization Method Generative Diversity

December 22, 2023

Quality-Diversity Generative Sampling for Learning with Synthetic Data
Allen Chang, Matthew C. Fontaine, Serena Booth, Maja J. Matarić, Stefanos Nikolaidis
Generative Model Synthetic Data Generative Modeling Synthetic Training Generative Diversity

August 25, 2023

Integrating LLMs and Decision Transformers for Language Grounded Generative Quality-Diversity
Achkan Salehi, Stephane Doncieux
Reinforcement Learning Medical LLM Human Language Decision Transformer Generative Agent Behavior Space Behavior Descriptor Generative Diversity

July 12, 2023

Diversity-enhancing Generative Network for Few-shot Hypothesis Adaptation
Ruijiang Dong, Feng Liu, Haoang Chi, Tongliang Liu, Mingming Gong, Gang Niu, Masashi Sugiyama, Bo Han
Unlabeled Data Generative Network Source Domain Shot Adaptation Generative Diversity

July 3, 2023

VOLTA: Improving Generative Diversity by Variational Mutual Information Maximizing Autoencoder
Yueen Ma, Dafeng Chi, Jingjing Li, Kai Song, Yuzheng Zhuang, Irwin King
Text Generation Variational Autoencoder Generative Quality Adversarial Code Bolt Rotation Generative Diversity

October 16, 2022

DiffGAR: Model-Agnostic Restoration from Generative Artifacts Using Image-to-Image Diffusion Models
Yueqin Yin, Lianghua Huang, Yu Liu, Kaiqi Huang
Image Generation Image Restoration Image to Image Diffusion Artifact Reduction Generative Diversity

June 14, 2022

Towards Goal, Feasibility, and Diversity-Oriented Deep Generative Models in Design
Lyle Regenwetter, Faez Ahmed
Generative Model Product Design Task Feasibility Pseudo Goal Generative Diversity

December 13, 2021

GM Score: Incorporating inter-class and intra-class generator diversity, discriminability of disentangled representation, and sample fidelity for evaluating GANs
Harshvardhan GM, Aanchal Sahu, Mahendra Kumar Gourisaria
Generative Model Generative Adversarial Network GAN Model Discriminative Task GAN Algorithm Sample Quality Analytic Score Generative Diversity

Generative Diversity

Papers

DiverGen: Improving Instance Segmentation by Learning Wider Data Distribution with More Diverse Generative Data

DE-CGAN: Boosting rTMS Treatment Prediction with Diversity Enhancing Conditional Generative Adversarial Networks

Pick-and-Draw: Training-free Semantic Guidance for Text-to-Image Personalization

Quality-Diversity Generative Sampling for Learning with Synthetic Data

Integrating LLMs and Decision Transformers for Language Grounded Generative Quality-Diversity

Diversity-enhancing Generative Network for Few-shot Hypothesis Adaptation

VOLTA: Improving Generative Diversity by Variational Mutual Information Maximizing Autoencoder

DiffGAR: Model-Agnostic Restoration from Generative Artifacts Using Image-to-Image Diffusion Models

Towards Goal, Feasibility, and Diversity-Oriented Deep Generative Models in Design

GM Score: Incorporating inter-class and intra-class generator diversity, discriminability of disentangled representation, and sample fidelity for evaluating GANs