Generative model attacks CNN on MNIST by subtly replacing input patterns.
problem Adversarial attacks on neural networks.
method Generative model that replaces input patterns with generated ones.
result Demonstrated effectiveness on MNIST dataset.
On-device research index
arXiv research
A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.
169,181 papers · 148 categories
Trend · papers per month
48 results for “MNIST replacement”
Fashion-MNIST replaces MNIST for machine learning benchmarks.
problem No new problem introduced.
method No new method introduced.
result Fashion-MNIST serves as a direct replacement for MNIST.
Reconstructs lost MNIST dataset to validate previous findings.
problem Lost precise derivation steps of MNIST dataset.
method Trace each MNIST digit to NIST source, reconstruct complete test set.
result Reconstructed MNIST dataset confirms previous findings on classifier performance.
FIS-GAN uses importance sampling in GANs to speed up training.
problem Efficiency in GAN training by focusing on hard-to-generate examples.
method Adapting importance sampling into GANs using normalizing flows.
result Significant acceleration in GAN optimization with improved fidelity.
New handwritten digits dataset for Kannada script.
problem Lack of datasets for Kannada numeral digits.
method Developed Kannada-MNIST and Dig-MNIST datasets.
result Initial CNN accuracy is lower than MNIST, indicating a challenge in generalization.
Improved image classification using centroids and stochastic sampling.
problem Limited accuracy of nearest-neighbor classification.
method Coarse-graining (replacing images by centroids) and stochastic sampling of centroids.
result Stochastic sampling of centroids improves classification accuracy.
Quantum-assisted GAN learns MNIST and LSUN datasets.
problem Learning latent variable generative models with adversarial networks.
method Generative adversarial learning with quantum annealing.
result Quantum-assisted GAN successfully learns MNIST and LSUN datasets.
Proposes a method to update model weights by dynamically changing features during training.
problem Updating model weights efficiently during training.
method Dynamic feature replacement during training to update model weights.
result Increased model accuracy with same training time and complexity.
Paper proposes a method to evaluate generative models using classifier accuracy.
problem Difficulty in objectively evaluating generative models.
method Train a classifier on generated data and compare its accuracy to real images.
result Generative models from VAE and GAN frameworks perform best on MNIST and fashion MNIST datasets.
Recently, fully-connected and convolutional neural networks have been trained to achieve state-of-the-art performance on a wide variety of tasks such as speech recognition, image classification, natural language processing, and bioinformatics. For classification tasks, most of these "deep learning" models employ the so…
New SSIM measure improves adversarial attacks' perceptual quality.
problem Lack of perceptual quality in traditional adversarial attacks.
method Replaced norms with SSIM measure for better quality.
result SSIM-constrained attacks achieve better quality and similar success rate.
Federated learning uses worst-case optimization to handle uncertain local data impacts.
problem Handling uncertainty in local data sets in federated learning.
method Reformulate FL problem using worst-case optimization theory, considering local data as uncertain functions bounded in a closed region.
result Comparison of FL performance with centralized learning and application of regularization factors.
Network deconvolution removes redundant data to improve neural network performance.
problem Redundant data in neural networks makes training challenging.
method Network deconvolution optimally removes pixel-wise and channel-wise correlations before each layer.
result Network deconvolution improves performance across various datasets.
The Tree-SNE Tree Existsstat.ML
A new method, tree-SNE, solves the scale problem in t-SNE.
problem Clustering and visualizing high-dimensional data, especially MNIST digits.
method Revisits t-SNE idea to create a 2+1 dimensional embedding with a scale parameter.
result The optimal embedding depends continuously on the scale parameter for all initial conditions.
Functional transfer matrices replace weights in neural networks, achieving high accuracy.
problem Representing connections in neural networks with functions instead of weights.
method Developed functional transfer matrices, stacked them with bias vectors and activations, and trained them using back-propagation.
result Deep functional transfer neural networks can be trained to achieve high test accuracies on the MNIST database.
Improved RNNs with flexible gates using kernel activation functions.
problem Modeling long-term dependencies in sequential data.
method Designed a more flexible architecture with adaptable parameters using kernel activation functions.
result Improved accuracy with negligible computational cost and speed-up in training iterations.
FSPool improves set prediction accuracy and convergence.
problem Set prediction models struggle with simple datasets due to the responsibility problem.
method Featurewise sort pooling to construct a permutation-equivariant auto-encoder.
result FSPool improves reconstructions and representations on various datasets.
Adversarial training improves model robustness with Voronoi constraints.
problem Adversarial examples mislead machine learning models, leading to incorrect classifications.
method Geometric framework using Voronoi cells to constrain adversarial training.
result Adversarial training with Voronoi constraints produces robust models.
New objective function reduces posterior collapse in generative models.
problem Posterior collapse in generative models with small datasets and high latent dimensions.
method Replaces ELBO's KL divergence with MMD, introduces latent clipping.
result μ-VAE outperforms ELBO and β-VAE models in quality and stability.
A new capsule routing algorithm derived from Variational Bayes improves performance in neural networks.
problem Improving performance of capsule networks in recognizing objects and their parts.
method Proposed a new capsule routing algorithm derived from Variational Bayes to fit a mixture of transforming gaussians.
result Significant improvement in MNIST to affNIST generalization over previous works.
Generative moment matching network (GMMN) is a deep generative model that differs from Generative Adversarial Network (GAN) by replacing the discriminator in GAN with a two-sample test based on kernel maximum mean discrepancy (MMD). Although some theoretical guarantees of MMD have been studied, the empirical performanc…
Pseudorandom Streams within Diffusion Models Act as Learnable Inputs That Affect Generation Qualitycs.LG
Pseudorandom inputs in diffusion models affect generation quality.
problem Pseudorandom inputs in diffusion models can be learned and affect model performance.
method Used a small multilayer perceptron to predict next values in pseudorandom orbits and a diffusion probe to replace real images with random tensors.
result Pseudorandom inputs can produce markedly different diffusion losses and generation quality.
Deep Fried Convnetscs.LG
The fully connected layers of a deep convolutional neural network typically contain over 90% of the network parameters, and consume the majority of the memory required to store the network parameters. Reducing the number of parameters while preserving essentially the same predictive performance is critically important …
Integrates differentiable decision trees into neural networks for faster training and inference.
problem Combining differentiability and conditional computation in tree ensembles for neural networks.
method Sparse activation function and specialized forward/backward propagation algorithms for efficient training and inference.
result 10x speed-ups and 20x reduction in parameters compared to existing methods, while maintaining performance.
This paper removes near-duplicates from Fashion-MNIST to improve testing accuracy.
problem Near-duplicate images in Fashion-MNIST increase testing accuracy, reducing dataset quality.
method Identified and removed near-duplicate images between training and testing sets.
result Improved dataset quality for better testing accuracy in machine learning models.
NAF combines neural networks with autoregressive models for better density estimation.
problem Improving density estimation and speech synthesis speed.
method Generalizes autoregressive models using neural networks for invertible transformations.
result NAF is a universal approximator for continuous probability distributions and outperforms IAF.
WTM reduces clause usage and computation time for pattern recognition.
problem High computation time and memory usage in Tsetlin Machine.
method Weighting clauses and using binomial sampling to reduce complexity.
result WTM achieves similar accuracy with fewer clauses and faster training.
A new framework for flexible neural network receptive fields.
problem Adaptive and flexible receptive fields in neural networks.
method Density-embedding layers that replace affine transformations with scalar products of input and density functions.
result Density-embedding layers can adaptively tune receptive fields and are computationally efficient.
New neural network model with rectified edges for classification.
problem Classifying data with monotone properties.
method Introduces rectified wire networks with a novel learning algorithm.
result Demonstrates SDA algorithm achieving zero loss for correct classification.
A training-free method for conditional sampling using flow matching.
problem Weight degeneracy in high-dimensional importance sampling.
method Sequential Monte Carlo with resampling and stochastic flow.
result Significantly outperforms existing methods on MNIST and CIFAR-10.
We introduce a new activation function using Chebyshev-Lagrange polynomials for improved neural network performance.
problem Improving data efficiency and accuracy of neural networks.
method Parameterized piece-wise polynomial activation functions based on Chebyshev nodes and Lagrangian interpolation.
result Significant improvements in model capacity and accuracy, especially in linear extrapolation.
MNIST-NET10 fusion improves MNIST classification to 0.1% error rate.
problem Improving MNIST classification accuracy.
method Complex heterogeneous fusion architecture using degree of certainty aggregation.
result MNIST-NET10 achieves 0.1% error rate with 10 misclassifications.
MNIST-Nd offers synthetic datasets to benchmark clustering across dimensions.
problem Clustering performance degrades with high-dimensional data.
method Training mixture variational autoencoders on MNIST to create synthetic datasets with varying latent dimensions.
result Leiden clustering algorithm is most robust as dimensionality grows.
TzK model learns tight conditional priors using side information.
problem Learning tight conditional priors using side information.
method TzK is a conditional probability flow-based model that exploits attributes to learn tight conditional priors around target observations. It trains via approximated ML and supports supervised, unsupervised, and semi-supervised learning.
result TzK produces efficient and stable approximations of arbitrary data distributions, comparable to state-of-the-art models.
Translating or rotating an input image should not affect the results of many computer vision tasks. Convolutional neural networks (CNNs) are already translation equivariant: input image translations produce proportionate feature map translations. This is not the case for rotations. Global rotation equivariance is typic…
FMP sampling improves model calibration without sharing data.
problem Difficulties in specifying priors for modern neural networks.
method FMP sampling replaces prior and likelihood with a predictive distribution, running the sampler centrally.
result FMP sampling significantly improves calibration over baseline methods.
Morpho-MNIST benchmarks unsupervised learning representations.
problem Lack of suitable benchmarks for evaluating learned representations.
method Extended MNIST dataset with morphometric analysis and proposed perturbations.
result Models can be quantitatively compared and characterised.
Poisson learning improves graph-based semi-supervised learning at very low label rates.
problem Degeneracy of Laplacian semi-supervised learning at low label rates.
method Replaces label assignment with source and sink placement, solving Poisson equation.
result Provably more stable and informative predictions than Laplacian learning.
q-CNN learns data features through entangled states.
problem Classifying MNIST and Fashion MNIST datasets.
method Introduces q-CNN model, a tensor network description, and studies entanglement structure.
result q-CNN learns entanglement structure to perform classification tasks accurately.
Improved classifier accuracy by using more of the class-specific structure in trained models.
problem Softmax ignores valuable information encoded in the full array of class response distributions.
method Developed a hybrid classifier (Softmax-Pooling Hybrid, ) that uses Softmax on high-scoring samples and a log-likelihood method on low-scoring samples.
result Reduces test set error by 6% to 23% using the exact same trained model.
New bounds for shallow neural networks with deterministic parameters.
problem Developing generalisation bounds for shallow neural networks.
method PAC-Bayesian theory applied to shallow neural networks with deterministic parameters.
result Empirical non-vacuous bounds for shallow neural networks trained with vanilla SGD.
Investigates the impact of batch size on GPU and TPU performance.
problem Optimizing performance of GPUs and TPUs during training and inference phases.
method Investigated the impact of batch size on performance of GPUs and TPUs using standard MNIST and Fashion-MNIST datasets.
result Significant speedup was achieved even with low-scale usage of TPUv2 units, up to 10x for training and 2x for prediction.
Bidirectional VAE reduces parameters and improves image tasks.
problem Improving image reconstruction, classification, interpolation, and generation.
method Uses a single neural network for both encoding and decoding in both forward and backward directions.
result Bidirectional VAEs reduce parameters by almost 50% and slightly outperform unidirectional VAEs.
A high-parallelism SNN improves feature learning efficiency and robustness.
problem Slow learning speed and limited learning capability in existing SNNs.
method Inspired by Inception modules, high-parallelism architecture, Vote-for-All decoding, adaptive repolarization mechanism.
result Superior performance and competitive accuracy compared to state-of-the-art unsupervised SNNs.
InfoMax Autoencoder maximizes mutual information for robust data representation.
problem Learning robust data representations from raw data.
method Explicitly maximizes mutual information between input and hidden representation.
result IMAE outperforms state-of-the-art models in clustering performance.
Quantum computing at room temperature achieves high accuracy in image classification.
problem Classifying images with single photons at room temperature.
method Optical transformation of quantum state to exploit interference and entanglement.
result Theoretical accuracy of 41.27% for MNIST and 36.14% for Fashion-MNIST.
Tensor methods improve image classification performance.
problem Improving image classification accuracy using tensor-based machine learning.
method Two novel tensor-based approaches: MANDy kernel reformulation and alternating ridge regression in tensor-train format.
result Tensor-based methods are competitive with state-of-the-art neural networks.
Ribbon: Scalable Approximation and Robust Uncertainty Quantification
problem Reliably quantifying predictive uncertainty for complex models
method Ribbon, a scalable approximation to Dirichlet-reweighted bootstrap uncertainty
result Asymptotically equivalent to a flat-prior Laplace approximation under correct likelihood specification, recovers robust sandwich covariance under misspecification