Searcharxiv⌕ Search

arXiv subjects

Zhiyong Gan

Publications and source records attributed to Zhiyong Gan.

5 recordsLinked to original sources

The minimum degree of minimally k-extendable graphs for some k

A graph on at least 2k+2 vertices is k-extendable if every matching of size k extends to a perfect matching, and it is minimally k-extendable if it is k-extendable but deleting any edge is not. Plummer proved that every k-extendable graph is (k+1)-connected. So its minimum degree is at least k+1. Lou and Yu conjectured that a minimally k-extendable graph on at most 4k vertices has minimum degree in {k+1,2k,2k+1}; the case k=1 and the claw-free case of k=2 have been settled previously. We determine the minimum degree of minimally k-extendable graphs for k=2 and k=3, thereby settling the Lou--Yu conjecture for these two values. For k=2 the minimum degree is in {3,4,5}. For k=3 it is in {4,7} at orders 8 and 10, and in {4,6,7} at order 12. The proofs pass to the complement graph and combine Tutte's perfect matching theorem with a classification of four-vertex and six-vertex obstructions. All predicted values are realized by explicit constructions.

math.CO↗

An extremal question of Erdős et al. on cycles

This paper addresses an extremal problem on long cycles proposed by Erdős, Faudree, Schelp and Simonovits in 1989. Let \(2\le k\le d<n\) be positive integers. For an n-vertex simple graph G, let \(θ_+^G\) count vertices of degree at least d, and \(C_{\ge k+1}\) denote a cycle of length at least \(k+1\). The original conjecture claimed that G contains a \(C_{\ge k+1}\) whenever \(θ_+^G \ge \lfloor k/2 \rfloor \lfloor (n-1)/d \rfloor+2\). Li, Ma and Ning disproved it with infinite counterexamples and gave linear lower bounds. Gan, Yi and Jiang later obtained piecewise sufficient conditions depending on the parity of k, whose bounds are sharp only when \(k=d\) and loose for \(k\ne d\). We derive a new sufficient condition for \(C_{\ge k+1}\). For odd \(k=d\), we use Woodall's sharp bound \(θ_+^G \ge k(n-2)/(2(k-1))+1\). For all other admissible parameters, our threshold is unified and independent of the parity of k: \(θ_+^G \ge \lfloor k/2 \rfloor (n-1)/d + (n-d-1)\lfloor k/2 \rfloor/(d(d-\lfloor k/2 \rfloor))+1\). We construct \(C_{\ge k+1}\)-free graphs to prove tightness for all admissible k. This resolves this long-standing Erdős-type extremal problem concerning high-degree vertices and long cycles.

math.CO↗

TTL: Test-time Textual Learning for OOD Detection with Pretrained Vision-Language Models

Vision-language models (VLMs) such as CLIP exhibit strong Out-of-distribution (OOD) detection capabilities by aligning visual and textual representations. Recent CLIP-based test-time adaptation methods further improve detection performance by incorporating external OOD labels. However, such labels are finite and fixed, while the real OOD semantic space is inherently open-ended. Consequently, fixed labels fail to represent the diverse and evolving OOD semantics encountered in test streams. To address this limitation, we introduce Test-time Textual Learning (TTL), a framework that dynamically learns OOD textual semantics from unlabeled test streams, without relying on external OOD labels. TTL updates learnable prompts using pseudo-labeled test samples to capture emerging OOD knowledge. To suppress noise introduced by pseudo-labels, we introduce an OOD knowledge purification strategy that selects reliable OOD samples for adaptation while suppressing noise. In addition, TTL maintains an OOD Textual Knowledge Bank that stores high-quality textual features, providing stable score calibration across batches. Extensive experiments on two standard benchmarks with nine OOD datasets demonstrate that TTL consistently achieves state-of-the-art performance, highlighting the value of textual adaptation for robust test-time OOD detection. Our code is available at https://github.com/figec/TTL.

cs.CL↗

ReplayCAD: Generative Diffusion Replay for Continual Anomaly Detection

Continual Anomaly Detection (CAD) enables anomaly detection models in learning new classes while preserving knowledge of historical classes. CAD faces two key challenges: catastrophic forgetting and segmentation of small anomalous regions. Existing CAD methods store image distributions or patch features to mitigate catastrophic forgetting, but they fail to preserve pixel-level detailed features for accurate segmentation. To overcome this limitation, we propose ReplayCAD, a novel diffusion-driven generative replay framework that replay high-quality historical data, thus effectively preserving pixel-level detailed features. Specifically, we compress historical data by searching for a class semantic embedding in the conditional space of the pre-trained diffusion model, which can guide the model to replay data with fine-grained pixel details, thus improving the segmentation performance. However, relying solely on semantic features results in limited spatial diversity. Hence, we further use spatial features to guide data compression, achieving precise control of sample space, thereby generating more diverse data. Our method achieves state-of-the-art performance in both classification and segmentation, with notable improvements in segmentation: 11.5% on VisA and 8.1% on MVTec. Our source code is available at https://github.com/HULEI7/ReplayCAD.

cs.CV↗

FodFoM: Fake Outlier Data by Foundation Models Creates Stronger Visual Out-of-Distribution Detector

Out-of-Distribution (OOD) detection is crucial when deploying machine learning models in open-world applications. The core challenge in OOD detection is mitigating the model's overconfidence on OOD data. While recent methods using auxiliary outlier datasets or synthesizing outlier features have shown promising OOD detection performance, they are limited due to costly data collection or simplified assumptions. In this paper, we propose a novel OOD detection framework FodFoM that innovatively combines multiple foundation models to generate two types of challenging fake outlier images for classifier training. The first type is based on BLIP-2's image captioning capability, CLIP's vision-language knowledge, and Stable Diffusion's image generation ability. Jointly utilizing these foundation models constructs fake outlier images which are semantically similar to but different from in-distribution (ID) images. For the second type, GroundingDINO's object detection ability is utilized to help construct pure background images by blurring foreground ID objects in ID images. The proposed framework can be flexibly combined with multiple existing OOD detection methods. Extensive empirical evaluations show that image classifiers with the help of constructed fake images can more accurately differentiate real OOD images from ID ones. New state-of-the-art OOD detection performance is achieved on multiple benchmarks. The code is available at \url{https://github.com/Cverchen/ACMMM2024-FodFoM}.

cs.CV↗