#image classification
13 papers match
A Fuzzy Rule-based Neuro-Symbolic Approach for Pipe Severity Prediction in Sewer Networks
Ngoc Thai Le, Thanh Ma, Umberto Straccia
The paper presents a modular neuro‑symbolic system that uses a Swin Transformer to predict multilabel pipe defect codes from images and then applies fuzzy IF‑THEN rules derived fro…
Kohn-Sham Spectral Embedding on Sparse Graphs at the Nishimori Temperature for Image Classification
V. S. Usatyuk, D. A. Sapozhnikov, S. I. Egorov
The paper proposes Kohn‑Sham Spectral Embedding (KSSE), a physics‑inspired, sparse‑graph spectral method that replaces dense CNN classifiers with a regularized Laplacian evaluated…
Contrastive Concept Importance: Explaining Pairwise Class Decisions Through Automatically Extracted Concept Representations
Roel Visser, Isaac Roberts, Barbara Hammer
The paper proposes Contrastive Concept Importance (CCI), a method that attributes the logit margin between a target and a foil class to automatically extracted visual concepts, pro…
Benign on Label, Malicious by Design: Clean-Label Dormant-to-Activated Backdoor via Machine Unlearning with Removable Camouflage
Dongdong Zhao, Can Li, Xiang Yao +3
The paper proposes a clean‑label backdoor attack that stays dormant during training and becomes active only after specific camouflage samples are removed via machine unlearning, us…
Improving the Robustness/Accuracy Tradeoff Against Adversarial Attacks Using Information Bottleneck Distillation Through Dual Teachers
Vincent Ryusuke Takahashi, Yoshinari Takeishi, Jun'ichi Takeuchi +1
The paper extends information bottleneck distillation by adding a clean‑trained teacher alongside a robust teacher, using cross‑layer attention to improve both clean accuracy and a…
Simplifying Neural Networks During Training
Lorenzo Sciandra, Samuele Fonio, Roberto Esposito
The paper proposes a training framework that monitors representation dynamics using the Inverse Fisher Criterion to identify when and where to replace later layers of a deep networ…
Representation Trajectories Matters: Complementary Evidence for OOD Detection and Image Classification
Ignacio M. De la Jara, Cristian Rodriguez-Opazo, Hamed Damirchi +2
The paper investigates how the step‑by‑step changes in a vision model’s internal representations (representation trajectories) can be used to improve out‑of‑distribution detection…
FunnelAL: Retrieve-then-Rank Active Learning for Single-Class Discovery
Reihaneh Rostami, Brian Goodwin
FunnelAL is an active learning system that first retrieves candidate images using embeddings and then ranks them to efficiently discover a single target class while minimizing anno…
Lightweight Image Classification of Raptor Species for Edge Devices: Rare-Species Dataset Expansion via Video Frame Extraction, Knowledge Distillation, and TensorRT Deployment
Takeshi Nishikawa
The paper presents a lightweight model ensemble for classifying raptor species on edge devices, using knowledge distillation from a large teacher model and expanding the dataset vi…
When Pretty Isn't Useful: Investigating Why Modern Text-to-Image Models Fail as Reliable Training Data Generators
Krzysztof Adamkiewicz, Brian Bernhard Moser, Stanislav Frolov +3
The paper evaluates modern text-to-image diffusion models as sources of synthetic training data and finds that, despite higher visual quality, newer models produce less diverse ima…
Screening Is Effective for Visual Recognition
Shunya Shimomura, Kazuhiro Hotta
The paper proposes VisionScreen, a model that applies a screening mechanism to evaluate and select relevant image patches independently, improving visual recognition performance co…
Gravitational lensing of gravitational waves: universal characteristics of strongly lensed memory waveforms
Ruanjing Zhang, Zhi-Chao Zhao, Shaoqi Hou +3
The paper analyzes how strong gravitational lensing alters the gravitational‑wave memory signal, revealing universal, image‑type‑dependent waveform features that are independent of…
Data Safety: Synthetic Data Quality Analysis Using CIFAKE Dataset
Kuniko Paxton, Amila AkagiÄ, Koorosh Aslansefat +2
The paper examines how synthetic images generated by different methods differ from real images in feature space, color statistics, and model training, and proposes strategies for e…
One search, two signals: results blend meaning (embedding similarity, so papers that never use your words still surface) with keyword matches on titles, abstracts and summaries. Free, no sign-in needed.