Arrow Research search

Author name cluster

Konstantin Sobolev

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

2 papers
1 author row

Possible papers

2

IS Journal 2026 Journal Article

Lightweight Attribute Localizing Models for Pedestrian Attribute Recognition

  • Ashish Jha
  • Dmitrii Ermilov
  • Anh Huy Phan
  • Konstantin Sobolev
  • Salman Ahmadi-Asl
  • Naveed Ahmed
  • Imran Junejo
  • Zaher Al Aghbari

Pedestrian attribute recognition (PAR) focuses on identifying attributes in pedestrian images, with applications in person retrieval, suspect reidentification, and soft biometrics. However, neural networks for PAR suffer from overparameterization and high computational complexity, making them unsuitable for resource_constrained devices. Tensor_based compression methods factorize layers without preserving the gradient direction during compression, leading to inefficient compression and an accuracy loss. We propose a novel approach for determining the optimal ranks of low_rank layers, ensuring that the gradient direction of the compressed model aligns with that of the original model. This means that the compressed model preserves the update direction of the full model, enabling more efficient compression for PAR tasks. The proposed procedure optimizes the compression ranks for each layer within the attribute localization model, followed by compression using canonical polyadic decomposition with error_preserving correction or singular value decomposition. This results in a reduction in model complexity while maintaining high performance.

Details DOI

AAAI Conference 2026 Conference Paper

T-LoRA: Single Image Diffusion Model Customization Without Overfitting

  • Vera Soboleva
  • Aibek Alanov
  • Andrey Kuznetsov
  • Konstantin Sobolev

While diffusion model fine-tuning offers a powerful approach for customizing pre-trained models to generate specific objects, it frequently suffers from overfitting when training samples are limited, compromising both generalization capability and output diversity. This paper tackles the challenging yet most impactful task of adapting a diffusion model using just a single concept image, as single-image customization holds the greatest practical potential. We introduce T-LoRA, a Timestep-Dependent Low-Rank Adaptation framework specifically designed for diffusion model personalization. In our work we show that higher diffusion timesteps are more prone to overfitting than lower ones, necessitating a timestep-sensitive fine-tuning strategy. T-LoRA incorporates two key innovations: (1) a dynamic fine-tuning strategy that adjusts rank-constrained updates based on diffusion timesteps, and (2) a weight parametrization technique that ensures independence between adapter components through orthogonal initialization. Extensive experiments show that T-LoRA and its individual components outperform standard LoRA and other diffusion model personalization techniques. They achieve a superior balance between concept fidelity and text alignment, highlighting the potential of T-LoRA in data-limited and resource-constrained scenarios.

PDF Details DOI