paper-with-me

Papers

Uncovering the Inner Workings of STEGO for Safe Unsupervised Semantic Segmentation

2023-04-14 · Alexander Koenig, Maximilian Schambach, Johannes Otterbach

Self-supervised pre-training strategies have recently shown impressive results for training general-purpose feature extraction backbones in computer vision. In combination with the Vision Transformer architecture, the DINO self-distillation technique has interesting emerging properties, such as unsupervised clustering in the latent space and semantic correspondences of the produced features without using explicit human-annotated labels. The STEGO method for unsupervised semantic segmentation contrastively distills feature correspondences of a DINO-pre-trained Vision Transformer and recently set a new state of the art. However, the detailed workings of STEGO have yet to be disentangled, preventing its usage in safety-critical applications. This paper provides a deeper understanding of the STEGO architecture and training strategy by conducting studies that uncover the working mechanisms behind STEGO, reproduce and extend its experimental validation, and investigate the ability of STEGO to transfer to different datasets. Results demonstrate that the STEGO architecture can be interpreted as a semantics-preserving dimensionality reduction technique.

📄 PDF Abstract BibTeX arXiv:2304.07314

Code (1)

merantix-momentum/stego-studies 공식 구현 pytorch

Tasks

Dimensionality ReductionSemantic SegmentationUnsupervised Semantic Segmentation

Methods 이 논문이 사용한 방법론

Attention 설명 없음
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
Multi-Head Attention 설명 없음
Absolute Position Encodings Absolute Position Encodings are a type of position embeddings for [Transformer-based models] where positional encodings are…
Position-Wise Feed-Forward Layer 설명 없음
Label Smoothing Label Smoothing is a regularization technique that introduces noise for the labels. This accounts for the fact that datasets may have mistakes in them, so maximizing the…
Dropout Dropout is a regularization technique for neural networks that drops a unit (along with connections) at training time with a specified probability $p$ (a common value is…
Residual Connection 설명 없음

Similar Papers 제목 키워드 기반

A Primer on the Inner Workings of Transformer-based Language Models

2024-04-30 · Javier Ferrando, Gabriele Sarti, Arianna Bisazza, Marta R. Costa-jussà

The rapid progress of research aimed at interpreting the inner workings of advanced language models has highlighted a need for contextualizing the insights gained from years of work in this area. This primer provides a c…

Decoder

Co-Activation Graph Analysis of Safety-Verified and Explainable Deep Reinforcement Learning Policies

2025-01-06 · Dennis Gross, Helge Spieker

Deep reinforcement learning (RL) policies can demonstrate unsafe behaviors and are challenging to interpret. To address these challenges, we combine RL policy model checking--a technique for determining whether RL polici…

Decision MakingDeep Reinforcement LearningReinforcement Learning (RL)Sequential Decision Making

Uncovering Unique Concept Vectors through Latent Space Decomposition

2023-07-13 · Mara Graziani, Laura O' Mahony, An-phi Nguyen, Henning Müller 외

Interpreting the inner workings of deep learning models is crucial for establishing trust and ensuring model safety. Concept-based explanations have emerged as a superior approach that is more interpretable than feature …

LLM Factoscope: Uncovering LLMs' Factual Discernment through Inner States Analysis

2023-12-27 · Jinwen He, Yujia Gong, Kai Chen, Zijin Lin 외

Large Language Models (LLMs) have revolutionized various domains with extensive knowledge and creative capabilities. However, a critical issue with LLMs is their tendency to produce outputs that diverge from factual real…

Inside the Visual Mind: Neuroscience-Motivated Concept Circuits for Interpreting and Steering Vision Transformers

2026-06-04 · Tang Li, Yanlin Chen, Mengmeng Ma, Xi Peng arxiv

Despite high accuracy, Vision Transformer (ViT) predictions can be driven by spurious cues, raising the need to understand their inner workings before safe deployment. Sparse autoencoders (SAEs) provide a promising lens …