Controlled training data generation with diffusion models

In this work, we present a method to control a text-to-image generative model to produce training data specifically “useful” for supervised learning. Unlike previous works that employ an open-loop approach and pre-define prompts to generate new data using either a language model or human expertise, we develop an automated closed-loop system which involves two feedback mechanisms. The first mechanism uses feedback from a given supervised model and finds adversarial prompts that result in image generations that maximize the model loss. While these adversarial prompts result in diverse data informed by the model, they are not informed of the target distribution, which can be inefficient. Therefore, we introduce the second feedback mechanism that guides the generation process towards a certain target distribution. We call the method combining these two mechanisms Guided Adversarial Prompts. We perform our evaluations on different tasks, datasets and architectures, with different types of distribution shifts (spuriously correlated data, unseen domains) and demonstrate the efficiency of the proposed feedback mechanisms compared to open-loop approaches. We also propose a text-to-image augmentation pipeline that modifies objects within semantic segmentation maps to enhance generalizability against distribution shifts in label space.

Name thesis:degree_name

Master of Science

Discipline thesis:degree_discipline

Computer Science

Grantor

The University of Texas at Austin

Year dc:date.issued

2024

Author dc:creator

• Ray, Ruchira

Advisors dc:contributor.advisor

• Krähenbühl, Philipp
• Zamir, Amir R.

Committee members dc:contributor.committeemember

• Roberto Martín-Martín
• Samantha Shorey

• Adversarial optimization
• Data augmentation
• Data generation
• Diffusion

Language dc:language.iso

English

Identifier URI

https://doi.org/10.26153/tsw/53429

OAI identifier oai:identifier

oai:repositories.lib.utexas.edu:2152/126892