A new VAE model learns global image representations by discarding texture.
problem Learning global image representations that ignore fine details.
method Combining VAE with autoregressive models to control global latent code.
result Improved generative modeling performance on image datasets.
WaiT improves image generation quality by focusing on natural frequency hierarchy.
problem Standard flow matching treats all spatial frequencies equally, ignoring natural frequency hierarchy.
method WaiT uses lossless wavelets to decompose generation into coarse and fine bands, waiting for the signal in high-frequency bands.
result WaiT achieves a pixel-space FID of 1.43 on ImageNet 512x512, reducing sampling compute by up to 50%.
Generative model creates user-specified textures from datasets.
problem Creating detailed textures from raw data.
method Generative adversarial networks with user control and adversarial loss.
result Model generates descriptive texture manifolds and 3D textures.
Paper introduces PSGAN for texture synthesis with manifold properties.
problem Texture synthesis with limited ability to handle diverse data.
method Generative adversarial networks (GAN) with extended noise tensor structure.
result PSGAN can smoothly interpolate and generate novel textures.
GANosaic generates high-resolution mosaic images from texture data.
problem Creating smooth, high-resolution mosaic images from texture data.
method Optimization in latent noise space of a generative texture model.
result Generative mosaic images with high resolution and smooth transitions.
Spatial GAN improves texture synthesis quality and scalability.
problem Texture synthesis quality and scalability issues.
method Spatial Generative Adversarial Networks (SGAN) with extended noise distribution.
result High quality textures, scalability, real-time generation, and fusion of diverse images.
Paper proposes TBSD for efficient anomaly detection in textured images.
problem Challenges in anomaly detection for textured images, especially in manufacturing systems.
method Texture basis integrated smooth decomposition (TBSD) approach.
result TBSD surpasses benchmarks with less misidentification and superior performance.
A new histogram layer improves texture analysis performance.
problem Extracting features for texture analysis from local spatial regions.
method Directly computes local spatial distribution of features during backpropagation.
result Improves performance on three material/texture datasets.
GOTEX synthesizes textures by optimizing feature distributions using optimal transport.
problem Texture synthesis with control over feature distributions.
method Optimal transport for statistical feature distribution comparison, semi-dual formulation, minimax optimization.
result GOTEX produces high-quality textures with various features, outperforming state-of-the-art methods.
Improves CNN robustness by reducing texture bias.
problem CNNs' reliance on local texture over global shape.
method Inspired by human vision, InfoDrop decorrelates model output from local texture.
result Enhanced robustness across various scenarios.
We apply the spike-and-slab Restricted Boltzmann Machine (ssRBM) to texture modeling. The ssRBM with tiled-convolution weight sharing (TssRBM) achieves or surpasses the state-of-the-art on texture synthesis and inpainting by parametric models. We also develop a novel RBM model with a spike-and-slab visible layer and bi…
Few-shot image classification is improved by correcting CNNs' texture bias.
problem Few-shot image classification performance is hindered by CNNs' texture bias.
method Corrected CNNs' texture bias using a simpler method than state-of-the-art approaches.
result State-of-the-art performance on miniImageNet task achieved.
Improved texture synthesis using wavelet-based statistics with rectifier non-linearity.
problem Improving texture synthesis quality using wavelet representations.
method Proposes a family of statistics based on non-linear wavelet representations with a generalized rectifier non-linearity.
result Significantly improves visual quality of texture synthesis compared to classical wavelet-based models.
Currently, Markov-Gibbs random field (MGRF) image models which include high-order interactions are almost always built by modelling responses of a stack of local linear filters. Actual interaction structure is specified implicitly by the filter coefficients. In contrast, we learn an explicit high-order MGRF structure b…
Macrocanonical models generate textures matching input features.
problem Generating textures that match specific features.
method Use Gibbs measures and minimize a convex function for sampling.
result Macrocanonical models can be applied to real-valued images under certain conditions.
CNNs trained on ImageNet favor textures over shapes, but can learn shape-based recognition.
problem CNNs trained on ImageNet favor textures over shapes, leading to biased recognition.
method Evaluating CNNs and human observers on images with a texture-shape cue conflict.
result CNNs trained on ImageNet favor textures over shapes, but can learn shape-based recognition.
A new liquid crystalline texture is proposed using gnomonic projection of the Hopf fibration.
problem Creating bend-free textures in flat space from 3-sphere Hopf fibration.
method Geodesic-preserving gnomonic projection of the Hopf fibration.
result A new liquid crystalline phase with only splay and twist.
Model for directed synthesis of audio textures using multi-scale RNNs.
problem Challenges in modeling complex audio textures with traditional methods.
method Combining multi-scale RNNs with a conditioning strategy for user-directed synthesis.
result Demonstrated improved performance on various audio texture datasets.
EnhanceNet improves image quality through automated texture synthesis.
problem Traditional image super-resolution methods produce over-smoothed images.
method Automated texture synthesis in adversarial training setting.
result Achieves state-of-the-art results in image quality benchmarks.
System generates speech textures and converts voices using backpropagation.
problem Generating speech textures and voice conversion from limited data.
method Approximate inversion of speech recognition network, matching neuron activations.
result System can generate realistic speech babble and reconstruct voices with limited data.
The paper classifies soil texture using 1D CNNs on hyperspectral data.
problem Classifying soil texture from hyperspectral data.
method Developed and implemented three 1D CNNs: LucasCNN, LucasResNet, and LucasCoordConv.
result LucasCoordConv achieves the best performance in accuracy.
Machine learning and complexity-entropy methods estimate liquid crystal properties from textures.
problem Extracting physical properties from liquid crystal textures.
method Combining permutation entropy, statistical complexity, and machine learning.
result Significant precision in predicting physical properties of liquid crystals.
A new Bayesian model improves dynamic texture segmentation.
problem Automatic selection of DTs in video sequences.
method Joint Dirichlet process mixture and GDTM approach with VBEM and RTSS.
result The proposed algorithm outperforms previous methods in efficiency and accuracy.
Enhanced rotation prediction improves SSL models by capturing both shape and texture information.
problem Rotation prediction misses texture information, limiting model performance.
method Introduces image enhanced rotation prediction (IE-Rot) that combines rotation and image enhancement tasks.
result IE-Rot models outperform Rotation on various benchmarks.
The paper introduces metrics to quantify information discarding in DNNs.
problem Understanding how input information is discarded during neural network processing.
method Developed two entropy-based metrics to measure pixel-wise and reconstruction uncertainty.
result The metrics provide new insights into DNN performance and information processing efficiency.
This paper introduces a new method for SAR imagery region discrimination using geodesic distances.
problem Region discrimination in monopolarized SAR imagery.
method Geodesic distance between GI0 models. result Advantages of using geodesic distance over stochastic distances.
Improved CAD system for lung disease diagnosis using transfer learning.
problem Early diagnosis of interstitial lung diseases is challenging.
method Transfer learning from general texture classification databases to improve CNN performance.
result 2% absolute increase in performance of the proposed CNN.
Deep learning boosts micro-CT image resolution and texture recovery.
problem Compensating for image resolution trade-offs in micro-CT imaging.
method EDSRGAN trained on a diverse dataset of uCT images.
result EDSRGAN outperforms other methods in texture recovery and resolution.
Algorithm segments glandular structures in colon histology images for cancer grading.
problem Manual gland segmentation is time-consuming and risky for patients.
method Local intensity and texture features, Random Forest classifier, multilevel approach.
result Fast, accurate automatic gland segmentation for clinical use.
We investigate the use of Deep Neural Networks for the classification of image datasets where texture features are important for generating class-conditional discriminative representations. To this end, we first derive the size of the feature space for some standard textural features extracted from the input dataset an…
This work improves texture segmentation by automatically tuning hyperparameters for Total-Variation.
problem The challenge is to automatically select hyperparameters for Total-Variation texture segmentation.
method The approach involves extending Stein's unbiased gradient estimator to handle correlated Gaussian noise, leading to an automatic tuning method.
result The method provides an automatic way to select hyperparameters for Total-Variation texture segmentation.
AT-CNNs show improved shape recognition over texture recognition.
problem Understanding adversarial training's impact on CNNs' feature learning.
method Systematic qualitative and quantitative approaches to interpret AT-CNNs.
result Adversarial training reduces texture bias and improves shape recognition.
Enhances sound texture in CNN for better acoustic scene classification.
problem Limited understanding of how CNNs perceive audio scenes.
method Used Class Activation Mapping (CAM) to analyze log-Mel features and proposed edge enhancement using DoG and Sobel operators.
result Edge-enhanced log-Mel features improve CNN performance in acoustic scene classification.
Generative model creates realistic images with 3D understanding.
problem Lack of 3D understanding in existing image generation models.
method Disentangled 3D representation using shape, viewpoint, and texture.
result Generates more realistic images and enables 3D operations.
In this on-going work, I explore certain theoretical and empirical implications of data transformations under the PCA. In particular, I state and prove three theorems about PCA, which I paraphrase as follows: 1). PCA without discarding eigenvector rows is injective, but looses this injectivity when eigenvector rows are…
Theory of packing diabolic domains in liquid crystals.
problem Understanding the packing of diabolic domains in liquid crystals.
method Lorentz transformations and geometric analysis.
result Diabolic domains can lower the elastic energy of the system.
Synthesizes faces from facial features, invariant to pose and expression.
problem Creating realistic face images from facial features.
method Learning facial landmarks and textures from facial-recognition features, training on frontal, neutral-expression images.
result Generated images are invariant to lighting, pose, and expression.
A scattering transform defines a signal representation which is invariant to translations and Lipschitz continuous relatively to deformations. It is implemented with a non-linear convolution network that iterates over wavelet and modulus operators. Lipschitz continuity locally linearizes deformations. Complex classes o…
Paper presents a method for satellite imagery classification using texture features.
problem Object identification in urban areas using satellite imagery.
method Pixel-level study with various features (correlation, homogeneity, energy, contrast). Supervised classification using SVM and Naive Bayes.
result Naive Bayes outperforms SVM with an overall accuracy of 76%.
Quantitative CT predicts ILD patterns and prognosis.
problem Diagnosing and predicting prognosis of fibrosing ILD patterns.
method High-resolution CT texture features, TM model for classification and survival analysis.
result TM model outperforms histogram-based model in distinguishing UIP from non-UIP patterns and allows for survival group partitioning.
Deep learning extracts terrain texture covariates for geostatistical modeling.
problem Improving prediction accuracy in geostatistical modeling using terrain texture data.
method Deep learning approach to automatically derive optimal terrain texture covariates from SRTM 90m DEM.
result Deep learning-derived covariates have strong explanatory power (R-squared around 0.6) for geochemical data.
Surface parameterizations and registrations are important in computer graphics and imaging, where 1-1 correspondences between meshes are computed. In practice, surface maps are usually represented and stored as 3D coordinates each vertex is mapped to, which often requires lots of storage memory. This causes inconvenien…
Paper compares XGB and BPNN for music style classification.
problem Efficient music style classification using different methods.
method Feature extraction for timbral texture, rhythmic content, and pitch content; comparative evaluation of XGB and BPNN.
result XGB outperforms BPNN for small datasets in music classification.
Deep learning framework predicts surface texture parameters and their uncertainties.
problem Predicting surface texture parameters and their uncertainties from multi-instrument datasets.
method Reproducible deep learning framework using multi-instrument dataset, quantile and heteroscedastic heads for uncertainty modeling, and post-hoc conformal calibration.
result High fidelity predictions (R2: Ra 0.9824, Rz 0.9847, RONt 0.9918) and well-modelled uncertainty targets (Ra_uncert 0.9899, Rz_uncert 0.9955).
Bayesian method selects interacting regions in Markov models.
problem Estimating interacting regions in Markov Random Fields.
method Reversible Jump Monte Carlo Markov Chain algorithm with pseudoposteriors.
result Proposed method accurately selects interacting regions in simulations and real data.
A new privacy-preserving ML framework using random sampling.
problem Achieving differential privacy in client-distributed ML.
method Draw and Discard: random model sampling for scalability and privacy.
result Demonstrated improved model quality and privacy guarantees.
Researchers found that avoiding synthetic data generation prevents model collapse in machine learning.
problem Model collapse in machine learning where models degenerate over generations.
method Comparing discard and augment workflows, focusing on Linear Regression.
result Theoretical evidence shows that for Linear Regression, test risk is bounded by π²/6 of original data alone.
Counting and classifying blood cells is an important diagnostic tool in medicine. Support Vector Machines are increasingly popular and efficient and could replace artificial neural network systems. Here a method to classify blood cells is proposed using SVM. A set of statistics on images are implemented in C++. The MPE…