ResNet-18
PulseAugur coverage of ResNet-18 — every cluster mentioning ResNet-18 across labs, papers, and developer communities, ranked by signal.
11 day(s) with sentiment data
-
Hybrid learning method boosts CNN training efficiency and accuracy
Researchers have developed a novel hybrid approach to accelerate the distributed training of convolutional neural networks (CNNs). This method combines a dual-batch learning scheme, which uses both large and small batch…
-
New method profiles Hessian mass in machine learning models
Researchers have developed a new method for evaluating block-diagonal curvature models in machine learning. This approach measures the squared off-block fraction of the parameter Hessian and analyzes the within-group cu…
-
New Forward-Forward algorithm variants boost image recognition accuracy
Researchers have explored the Forward-Forward (FF) algorithm, which trains networks layer-by-layer using a local "goodness function." The study systematically investigated how the selectivity and activity shape of these…
-
New PAVE method accelerates DNN verification by 10x
Researchers have developed a new method called PAreto VErification (PAVE) to address the scalability challenges in formally verifying deep neural networks (DNNs). PAVE augments DNNs with early exits, allowing for the ve…
-
LLMs dynamically select aggregation rules in federated learning
Researchers have developed FedReason, a novel approach that utilizes large language models (LLMs) to dynamically select aggregation rules in federated learning. Instead of relying on fixed rules, FedReason employs an LL…
-
New HDPU moves neural network inference to the sensor
Researchers have developed a Hardwired Distributed Processing Unit (HDPU) capable of performing neural network inference directly on a sensor, freeing up the host system. This HDPU, implemented on an Artix-7 FPGA, can p…
-
New method predicts neural network performance for efficient model pruning
Researchers have developed a method to predict the future performance of neural network candidates during training, allowing for more efficient pruning of underperforming models. This approach uses "nowcasters" trained …
-
New IDSISA method improves machine unlearning efficiency and utility
Researchers have developed IDSISA, a novel method for efficient machine unlearning that aims to remove the influence of specific training data while maintaining model utility. Unlike full retraining, which is computatio…
-
Quantum ML models audited for fairness in medical imaging
Researchers have conducted a fairness audit on quantum machine learning (QML) models used for medical imaging, finding that while QML can achieve competitive performance, it may still exhibit disparities across sensitiv…
-
STAR-FER framework enhances low-resolution facial expression recognition
Researchers have developed STAR-FER, a novel framework designed to improve facial expression recognition (FER) in low-resolution images while minimizing computational overhead. This adaptive selective restoration approa…
-
New research tests VICReg's standard deviation hinge for embedding stability
Researchers have experimentally tested the claim that a standard deviation hinge in VICReg, a method for preventing collapse in embeddings, offers a safety margin compared to a variance hinge. Using ResNet-18 on CIFAR-1…
-
New personalized federated learning method HDC-FL shows strong performance
Researchers have introduced HDC-FL, a novel method for personalized federated learning that leverages client-specific classifier heads to guide representation sharing. This approach groups client descriptors and creates…
-
New LEAN technique enhances neural network efficiency and error tolerance
Researchers have developed a new technique called LEAN (Loss-Engineered Modulation) to improve the efficiency and error tolerance of neural networks. LEAN works by altering the distribution of weight sensitivity in trai…
-
Physics-inspired optimizers show period-two orbits beyond stability limit
Researchers have explored physics-inspired optimizers for deep learning, focusing on their behavior beyond the stability limit. They identified period-two orbits that can maintain stability if the optimizer's elasticity…
-
New training method enables efficient on-device training of binary and ternary AI models
Researchers have developed a new training method called \"$\Sigma\Delta$-Sampler\" that leverages Ising solvers to train binary and ternary AI models more efficiently. This approach embeds a stochastic sampling and anne…
-
Noisy Labels Leave Geometric Signature in Neural Networks, Study Finds
Researchers have developed a method using persistent homology to analyze how noisy labels affect neural network representations. Their experiments on image and text classification tasks, including CIFAR-10 and CIFAR-100…
-
New Dormant Spiking Neuron model boosts SNN efficiency and accuracy
Researchers have introduced the Dormant Spiking Neuron (DSN) model, a new mechanism designed to improve the energy efficiency of Spiking Neural Networks (SNNs). The DSN allows neurons to enter a dormant state when their…
-
New typhoon dataset V3 benchmarks AI models for multi-spectral analysis
This paper introduces the Digital Typhoon Dataset V3, an updated collection of nearly 50 years of typhoon satellite imagery designed to evaluate machine learning models. The new version includes multispectral data to su…
-
New research explores advanced quantization for edge AI accelerators
Two new research papers explore advanced quantization techniques for deep neural networks (DNNs) on edge devices. The first paper introduces ShiftQAT, a quantization-aware training flow for the MAX7800x family of CNN ac…
-
New defense method shields ML models from black-box query attacks
Researchers have developed a new defense mechanism called Gradient Confusion Defense (GCD) to protect machine learning models from black-box score-based query attacks. This method, which does not require retraining the …