ResNet18
PulseAugur coverage of ResNet18 — every cluster mentioning ResNet18 across labs, papers, and developer communities, ranked by signal.
8 day(s) with sentiment data
-
Classical SU(2) models outperform quantum circuits on vision tasks
A new research paper compares classical SU(2) models with variational quantum circuits (VQCs) on various vision benchmarks. The study found that quaternion-valued neural networks, a type of classical SU(2) model, perfor…
-
New CPDA Framework Enhances Unsupervised Time-Series Domain Adaptation
Researchers have introduced Class-Conditional Path Distribution Alignment (CPDA), a novel framework for unsupervised time-series domain adaptation. Unlike existing methods that align marginal feature distributions, CPDA…
-
New machine unlearning method minimizes collateral damage to similar data
Researchers have developed a new machine unlearning method that aims to minimize collateral damage to semantically similar data. This approach, called retain-aware localization, considers the importance of model paramet…
-
Quantization impacts deep learning model explanations, study finds
A new study published on arXiv investigates how post-training quantization (PTQ) affects the explainability of deep learning models. Researchers evaluated five common CNN architectures (VGG19, ResNet18, EfficientNet-B0,…
-
Laser speckle material classification improved by physics-aware data augmentation
Researchers have investigated how data augmentation techniques impact the performance of deep learning models in classifying laser speckle patterns for material identification. Their study, using ResNet18 and EfficientN…
-
Quantization impacts deep learning model explanations, study finds
A new study investigates the impact of post-training quantization (PTQ) on the explainability of deep learning models, specifically focusing on five Convolutional Neural Network (CNN) architectures. Researchers found th…
-
QScheduler algorithm enables adaptive on-device AI training on microcontrollers
Researchers have developed QScheduler, an adaptive algorithm designed to optimize on-device training for microcontrollers equipped with Neural Processing Units (NPUs). This method estimates gradients using only forward …
-
New hierarchical ensemble methods improve zebrafish phenotype classification
Researchers have developed and evaluated three hierarchical ensemble methods for classifying zebrafish phenotypes from embryo images. The study compared three backbone architectures: ResNet18, ViT, and ConvNeXt. ConvNeX…
-
NeuronSoup architecture evolves temporal graphs without backpropagation
Researchers have developed NeuronSoup, a novel neural computation architecture that deviates from traditional layer-by-layer processing. Instead, it utilizes asynchronous, delay-mediated signal propagation through a sha…
-
Fisher Rank Inflation: Memorization Signature in Noisy Deep Learning
Researchers have identified a phenomenon called Fisher Rank Inflation, which serves as a spectral signature for memorization in deep learning models trained with label noise. This inflation occurs when models transition…
-
New research tackles audio deepfake detection with advanced AI techniques
Researchers are developing advanced methods for detecting audio deepfakes, focusing on improving generalization and interpretability. One approach, SONAR, uses a frequency-guided framework to exploit high-frequency arti…
-
New AI method identifies wildlife by gait dynamics using video analysis
Researchers have developed a novel, automated video-based system for identifying individual wild animals by analyzing their gait dynamics. This method utilizes the Segment Anything Model 3 (SAM3) to create precise anima…
-
New LINet architecture enables continuous cross-modal learning in RGB-D scene classification
Researchers have introduced LINet, a novel Multi-Stream Neural Network (MSNN) designed for RGB-D scene classification. Unlike existing architectures that fuse features discretely, LINet employs a continuous integration …
-
Deep learning models achieve 97% accuracy in automated brain tumor detection
Researchers have developed a deep learning approach using Convolutional Neural Networks (CNNs) and Residual Networks (ResNets) to automate the detection of brain tumors in MRI images. The study applied transfer learning…
-
New LaryngealCT Dataset Benchmarks Deep Learning for Cancer Staging
Researchers have developed LaryngealCT, a new benchmark dataset for staging laryngeal cancer using deep learning models. The dataset comprises 1,029 CT scans aggregated from The Cancer Imaging Archive and has been used …
-
New research explores privacy techniques for computer vision systems
Two new research papers explore methods for enhancing privacy in computer vision systems. The first paper, "PrivacyBench," introduces a framework to evaluate combinations of privacy techniques, revealing that combining …
-
LLMs struggle with zero-shot ECG diagnosis, CNNs outperform
A comparative study evaluated the efficacy of zero-shot multimodal large language models (LLMs) against Convolutional Neural Network (CNN) based models for classifying 12-lead ECG images. While LLMs like GPT-5.2, GPT-4.…
-
New concolic testing method enhances Transformer robustness analysis
Researchers have developed a new concolic testing method for Transformer classifiers that uses SHAP estimates to prioritize path predicates based on their influence on the model's predictions. This approach, implemented…
-
New Finetuning Method Adapts DNNs for ReRAM In-Memory Computing
Researchers have developed a new finetuning method to adapt deep neural networks for deployment on ReRAM-based in-memory computing hardware. This approach addresses the challenges of I-V non-linearity and retention erro…
-
Vision Transformers reduce demographic bias in face anti-spoofing systems
A new study published on arXiv investigates the impact of Vision Transformer (ViT) architectures on demographic bias in face presentation attack detection (PAD) systems. The research compares ViTs against convolutional …