Enhancing Out-of-Distribution Detection in Machine Learning by Increasing Auxiliary Outlier Diversity
Improving the diversity of auxiliary outlier data used during training is crucial for enhancing the generalization ability of out-of-distribution (OOD) detection methods in machine learning.
Non-Linear Class-wise Invariant Sampling (NCIS) for Improved Out-of-Distribution Detection in Image Classifiers
Synthesizing high-quality, near out-of-distribution images using a novel method called NCIS, which leverages diffusion models and non-linear class-conditional distributions, significantly improves a classifier's ability to detect out-of-distribution samples.
COOD: 개념 기반 제로샷 다중 레이블 OOD(Out-of-Distribution) 탐지 프레임워크
COOD는 사전 훈련된 비전-언어 모델(VLM)을 활용하여 복잡한 다중 레이블 설정에서 추가적인 재훈련 없이 OOD 샘플을 효과적으로 탐지하는 제로샷 프레임워크입니다.
COOD: Zero-Shot Out-of-Distribution Detection for Multi-Label Images Using Concept Expansion with Vision-Language Models
COOD is a novel zero-shot framework that leverages pre-trained vision-language models and a concept-based label expansion strategy to effectively detect out-of-distribution (OOD) samples in complex, multi-label image datasets.
語義空間或共變量空間？探討 Out-of-Distribution Detection 中難解問題的研究
現有的 Out-of-Distribution Detection 基準測試設定存在缺陷，因為它們沒有區分語義空間和共變量空間的偏移，導致某些 OOD 樣本無法被基於 ID 數據訓練的分類器識別。
A Study on the Intractable Case of Out-of-Distribution Detection: Is it Semantic or Covariate Shift?
The current definition of out-of-distribution (OOD) detection is flawed, making certain OOD testing protocols intractable for post-hoc methods. This paper proposes a more precise definition of "semantic shift" based on the training data, introducing the concepts of "Semantic Space" and "Covariate Space" to clarify the limitations of post-hoc OOD detection and define a "Tractable OOD" setting.
A Unified Approach to Out-of-Distribution Detection in Deep Learning: Addressing Spurious Correlations and Fine-Grained Classification
This research paper introduces ASCOOD, a novel method for improving out-of-distribution (OOD) detection in deep learning models by addressing the challenges posed by spurious correlations in training data and the subtle differences in fine-grained classification tasks.
Diffusion-Based Layer-Wise Semantic Reconstruction for Unsupervised Out-of-Distribution Detection Using Latent Feature Diffusion Network (LFDN)
This research paper introduces a novel method for unsupervised out-of-distribution (OOD) detection in computer vision, leveraging the power of diffusion models for reconstructing semantic features extracted from multiple layers of image data.
Category-Extensible Out-of-Distribution Detection Using Hierarchical Context Descriptions Learned from Vision-Language Models
This paper proposes a novel method called CATEX, which leverages the power of vision-language models like CLIP to improve out-of-distribution (OOD) detection in image classification. CATEX introduces hierarchical context descriptions - perceptual and spurious contexts - to define precise category boundaries, enabling the model to better distinguish between in-distribution and OOD samples, even in category-extended scenarios.
사전 훈련된 단일 단계 객체 감지 모델의 Out-of-Distribution 데이터에 대한 강건성 분석
사전 훈련된 단일 단계 객체 감지 모델은 새로운 객체 범주에 대한 재훈련 없이도 알 수 없는 객체를 감지하는 데 있어 고유한 강건성을 보여줍니다.
© 2024 by Linnk AI