paper-with-me

홈 › Papers

DivRL: Disentangled Self-Similarity Rewards for Diverse Subject-Driven Generation

2026-06-22 · Qian Wang, Zhenyu Li, Abdelrahman Eldesokey, Peter Wonka arxiv

Subject-driven image generation faces an "Identity-Diversity Paradox", where strong identity preservation often leads to rigid and low-diversity outputs. We propose a post-training framework called DivRL that jointly optimizes identity consistency and structural diversity simultaneously by leveraging disentangled visual features from a robust similarity model. Specifically, we introduce a Negative Self-Similarity Measure (nSSM) to quantify structural diversity, and Visual Semantic Matching (VSM) to evaluate identity consistency. We propose an "Explore-and-Suppress" strategy that treats VSM as a gated constraint: the model freely explores structurally diverse configurations, and only samples that violate the identity threshold are penalized via a quadratic hinge loss. This converts identity preservation from a competing objective into a feasibility constraint, allowing nSSM and VSM to improve jointly. Experiments demonstrate that our method effectively pushes the model to generate both consistent and diverse images and improves structural diversity while maintaining comparable identity consistency through a gated optimization formulation.

📄 PDF Abstract BibTeX arXiv:2606.23950

Code (0)

등록된 구현이 없습니다.

Tasks

Image Generation

Similar Papers 제목 키워드 기반

Self-evolving Autoencoder Embedded Q-Network

2024-02-18 · J. Senthilnath, Bangjian Zhou, Zhen Wei Ng, Deeksha Aggarwal 외

In the realm of sequential decision-making tasks, the exploration capability of a reinforcement learning (RL) agent is paramount for achieving high rewards through interactions with the environment. To enhance this cruci…

Decision MakingReinforcement Learning (RL)Sequential Decision Making

Disentangled and Self-Explainable Node Representation Learning

2024-10-28 · Simone Piaggesi, André Panisson, Megha Khosla

Node representations, or embeddings, are low-dimensional vectors that capture node properties, typically learned through unsupervised structural similarity objectives or supervised tasks. While recent efforts have focuse…

DisentanglementRepresentation Learning

RD$^2$: Reward Decomposition with Representation Decomposition

2020-12-01 · NeurIPS 2020 12 · Zichuan Lin, Derek Yang, Li Zhao, Tao Qin 외

Reward decomposition, which aims to decompose the full reward into multiple sub-rewards, has been proven beneficial for improving sample efficiency in reinforcement learning. Existing works on discovering reward decompos…

Zero-shot text-to-speech synthesis conditioned using self-supervised speech representation model

2023-04-24 · Kenichi Fujita, Takanori Ashihara, Hiroki Kanagawa, Takafumi Moriya 외

This paper proposes a zero-shot text-to-speech (TTS) conditioned by a self-supervised speech-representation model acquired through self-supervised learning (SSL). Conventional methods with embedding vectors from x-vector…

RhythmSelf-Supervised LearningSpeech Synthesistext-to-speech+2

Enhancing the Stability of LLM-based Speech Generation Systems through Self-Supervised Representations

2024-02-05 · Álvaro Martín-Cortinas, Daniel Sáez-Trigueros, Iván Vallés-Pérez, Biel Tura-Vecino 외

Large Language Models (LLMs) are one of the most promising technologies for the next era of speech generation systems, due to their scalability and in-context learning capabilities. Nevertheless, they suffer from multipl…

DecoderIn-Context Learningtext-to-speechText to Speech+1