MNIST database
PulseAugur coverage of MNIST database — every cluster mentioning MNIST database across labs, papers, and developer communities, ranked by signal.
- instance of CIFAR-10 90%
- instance of Fashion-MNIST 90%
- used by Fashion-MNIST 80%
- used by Quantum Machine Learning 80%
- used by CIFAR-10 70%
- used by CIFAR-100 70%
- used by federated learning 70%
- used by alphaXiv 70%
- used by Gotit.pub 70%
- used by ScienceCast 70%
- used by rectifier 70%
- used by graphics processing unit 70%
22 day(s) with sentiment data
-
Physical noise in quantum neural networks explored as a native regularizer
Researchers have explored the potential of using physical noise in photonic hybrid quantum neural networks (PHQNNs) as a native regularizer, drawing parallels to noise-injection techniques in classical deep learning. By…
-
New ECRAM system accelerates edge continual learning, slashing energy use
Researchers have developed CLASP, a novel system designed to accelerate continual learning on edge devices by integrating in-memory computing (IMC) with a specialized ECRAM device. This approach addresses the significan…
-
New methods enhance differential privacy in deep neural network training · 2 sources tracked
Two new research papers propose novel methods for training deep neural networks with differential privacy, aiming to improve both accuracy and efficiency. The first paper introduces an end-to-end framework that privatiz…
-
New framework estimates federated learning complexity for edge AI
Researchers have developed a new framework to estimate the learning complexity of federated perception systems before deployment. This classifier-agnostic approach combines intrinsic data properties like dimensionality …
-
New 'learnable novelty' metric unifies intelligence metrics
Researchers have introduced a new concept called "learnable novelty" as a unified measure for intelligence across various fields like statistics, computation, and agent behavior. This concept is quantified using a diffe…
-
SechKAN: New Neural Network Architecture Uses Hyperbolic Secant Functions
Researchers have introduced SechKAN, a novel neural network architecture that utilizes hyperbolic secant functions. This design aims to leverage the smooth, localized properties of the sech function for improved perform…
-
Decafs model improves generative AI interpretability and performance
Researchers have developed Decafs, a novel conditional generator based on Lie groups designed to improve the interpretability of flow-based generative models. By disentangling generative factors in the latent space thro…
-
New research tackles wireless federated learning challenges
Two new research papers explore advancements in federated learning for wireless environments. The first paper introduces a convergence-latency-aware adaptive modulation and resource allocation scheme for RIS-assisted wi…
-
New Orthogonal Gradient Method Explores Noisy-Label Memorization in Neural Networks
A new research paper introduces OrthoGrad, a method that modifies the optimizer's update by removing the component of each weight gradient parallel to the current weight vector. This geometric intervention was tested on…
-
New Conditioned Direct Feedback Alignment Method Improves Neural Network Training
Researchers have developed a new method called Conditioned Direct Feedback Alignment (nDFA) that improves the training of deep neural networks. This approach addresses a failure mode in Direct Feedback Alignment (DFA) b…
-
Boundary-seeking distillation fails in generative AI, study finds
Researchers have identified a fundamental flaw in boundary-seeking distillation techniques when applied to bottlenecked generative architectures. Through experiments on the MNIST dataset, they demonstrated that methods …
-
Sakana AI bypasses backpropagation with novel Error Diffusion training
Researchers at Sakana AI have developed a new deep learning training method called Error Diffusion (ED) that bypasses the need for backpropagation. This novel approach adheres to Dale's principle, a biological constrain…
-
Photonic CNN achieves high accuracy and energy efficiency
Researchers have developed a fully photonic convolutional neural network (PCNN) capable of executing image classification tasks entirely within the optical domain. This novel architecture integrates convolution, max-poo…
-
Research paper proposes synthetic data verification to prevent model collapse
A new research paper explores the phenomenon of "model collapse," where generative models trained on their own synthetic data degrade in performance over time. The study proposes that incorporating an external synthetic…
-
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…
-
New Langevin Computing Method Enhances Reservoir Diversity and Readout
Researchers have developed a new method for Langevin computing, a form of computation that utilizes thermal fluctuations. This approach, detailed in a recent paper, introduces moment-resolved readout and reservoir diver…
-
New parsimonious models proposed for skewed matrix clustering
Researchers have introduced a new family of 256 parsimonious models for mixtures of skewed matrix variate bilinear factor analyzers, specifically addressing the skew t distribution. The proposed method aims to reduce ov…
-
New PUe Framework Enhances Learning with Biased Datasets · 2 sources tracked
Researchers have developed a new framework called PUe to enhance Positive-Unlabeled (PU) learning by addressing selection bias in real-world datasets. This framework, building on prior work by Bekker et al., introduces …
-
New protocol detects functional fingerprints in converged neural networks
Researchers have developed a protocol to detect donor-specific functional fingerprints in neural networks after they have converged, a phenomenon known as Neural Collapse. By applying an affine-correct alignment mapping…
-
New Differentiable Algorithm Learns Cognitive Maps from Images
Researchers have developed a new algorithm called gradCSCG, which is a fully differentiable module designed to enable end-to-end learning of interpretable cognitive maps from raw image sequences. This approach builds up…