Image classifiers are sensitive to small changes, affecting most images in a class.
problem Sensitivity of image classifiers to small perturbations.
method Demonstrated sensitivity for any classifier over images, showing that for most classes, a tiny perturbation can change the classification of a majority of images.
result Image classifiers are sensitive to small perturbations, affecting most images in a class.
New framework tackles DG under posterior drift, where optimal classifier varies by domain.
problem Generalizing from multiple domains with varying optimal classifiers.
method Decision-theoretic framework for DG under posterior drift.
result Optimal classifier can vary significantly across domains, challenging existing DG approaches.
Designing effective and efficient classifier for pattern analysis is a key problem in machine learning and computer vision. Many the solutions to the problem require to perform logic operations such as `and', `or', and `not'. Classification and regression tree (CART) include these operations explicitly. Other methods s…
This work uses image generation models to find vision model bugs.
problem Automatically discovering failures in vision models.
method Conditional text-to-image generation and captioning models.
result Demonstrated utility of large-scale generative models to find vision model bugs.
When learning a new concept, not all training examples may prove equally useful for training: some may have higher or lower training value than others. The goal of this paper is to bring to the attention of the vision community the following considerations: (1) some examples are better than others for training detector…
The paper proposes methods to predict classifier generalization with few labeled samples.
problem Measuring classifier generalization with limited labeled data.
method Analysis of generalization variability, transfer-based solutions in supervised, semi-supervised, and unsupervised settings.
result Simple measures correlate with classifier generalization and can predict it with confidence.
Computer vision models are unstable due to task symmetries and labelling issues.
problem Instability of computer vision models in classification tasks.
method Analysis of symmetries, categorical nature, and labelling issues.
result Instability is a necessary result of current computer vision formulation.
A family of algorithms for time series classification (TSC) involve running a sliding window across each series, discretising the window to form a word, forming a histogram of word counts over the dictionary, then constructing a classifier on the histograms. A recent evaluation of two of this type of algorithm, Bag of …
Develops a robust multiclass classification method for deep image classifiers.
problem Tackles data contamination and robustness to outliers in deep image classifiers.
method Uses Distributionally Robust Optimization (DRO) with Wasserstein metric ambiguity sets and regularized learning.
result Reduces test error rate by up to 83.5% and loss by up to 91.3% in image classification tasks.
Framework for efficient defect classification and inspection.
problem Adaptive defect classification and inspection from high volume data.
method Continual learning framework for dynamic classifier updates.
result Efficient storage and computational needs reduction.
CutMix training technique improves spatial locality in Vision Transformers.
problem Improving spatial locality in Vision Transformers trained from scratch.
method Comparison of Baseline and Modern training protocols on CIFAR-10, CIFAR-100, and Tiny-ImageNet.
result CutMix training component significantly reduces Mean Attention Distance (MAD) in early layers of Vision Transformers.
We introduce a learning framework called learning using privileged information (LUPI) to the computer vision field. We focus on the prototypical computer vision problem of teaching computers to recognize objects in images. We want the computers to be able to learn faster at the expense of providing extra information du…
We find ways to make physical signals misclassified by computer vision models.
problem Vulnerability of signal classifiers to adversarial perturbations in physical signals.
method Solving PDE-constrained optimization problems to construct imperceptible perturbations.
result Effective and physically realizable adversarial perturbations can be computed for machine learning models.
Typically a classifier trained on a given dataset (source domain) does not performs well if it is tested on data acquired in a different setting (target domain). This is the problem that domain adaptation (DA) tries to overcome and, while it is a well explored topic in computer vision, it is largely ignored in robotic …
Model interprets image classification using hierarchical prototypes.
problem Lack of hierarchical interpretation in vision models.
method Uses hierarchically organized prototypes to classify objects at each level of a taxonomy.
result Model interprets image classification at each level of the taxonomy.
Deep networks reveal colour opponent cells under retinal constraints.
problem Understanding colour vision in deep neural networks.
method Classifying cells in early layers of anatomically constrained convolutional neural networks.
result Emergence of single and double opponent cells in deep networks.
The study examines how much data is needed for generative and vision-language models to make reliable predictions.
problem Ensuring reliable predictions with low data for models used in medical decision support.
method Analyzes uniform convergence bounds for VLM-induced classifiers under low-dimensional semantic representations.
result Finite-sample uniform convergence bounds for accuracy and calibration functionals of VLM-induced classifiers.
Active search (AS) on graphs focuses on collecting certain labeled nodes (targets) given global knowledge of the network topology and its edge weights under a query budget. However, in most networks, nodes, topology and edge weights are all initially unknown. We introduce selective harvesting, a variant of AS where the…
Deep neural networks are vulnerable to adversarial attacks in time series classification.
problem Vulnerability of deep learning models to adversarial time series examples.
method Proposed adversarial attack mechanisms to add noise to input time series.
result Current state-of-the-art deep learning time series classifiers are vulnerable to adversarial attacks.
Fair Mixup improves fairness in classifiers by interpolating between groups.
problem Ensuring fairness in classifiers during training and evaluation.
method Fair Mixup uses interpolation of samples between groups to enforce fairness constraints.
result Fair Mixup ensures better generalization of fairness in various benchmarks.
This study automates blood cell classification using computer vision.
problem Automated detection and classification of white blood cells.
method Color-based segmentation, morphological processing, feature extraction, PCA, Unsupervised and Supervised learning algorithms, Deep Convolutional Neural Networks.
result Identification of robust algorithms for automated blood cell classification.
Paper introduces SPADE method to protect classifiers from OOD and adversarial samples.
problem Protecting classifiers from out-of-distribution and adversarial samples.
method SPADE method based on GEV model in latent space.
result Provable protection against OOD and adversarial samples.
Improved few-shot visual reasoning with image preprocessing.
problem Few-shot classifiers struggle with abstract visual reasoning tasks.
method Spectral feature removal to emphasize unique image parts.
result Combining spectral preprocessing with Relational Networks improves accuracy nearly 40%.
Develops counterfactual visual explanations to show how images could change to classify differently.
problem Creating understandable explanations for vision system predictions.
method Selects a distractor image and identifies spatial regions to modify for different classification.
result Users trained with counterfactual explanations perform better in fine-grained bird classification.
Simple methods boost sound event classifier accuracy by 2.5%.
problem Label noise in sound event classification datasets.
method Label smoothing regularization, mixup, and noise-robust loss functions.
result Simple methods can boost accuracy by up to 2.5%.
Automatic classification of trees using remotely sensed data has been a dream of many scientists and land use managers. Recently, Unmanned aerial vehicles (UAV) has been expected to be an easy-to-use, cost-effective tool for remote sensing of forests, and deep learning has attracted attention for its ability concerning…
CDAM improves attention maps for ViTs, making them more class-sensitive.
problem Existing attention maps in ViTs lack class sensitivity.
method Class-discriminative attention maps (CDAM) that scale attention scores by class relevance.
result CDAM provides more class-sensitive explanations than existing methods.
Denoised smoothing defends pretrained classifiers against adversarial attacks.
problem Adversarial attacks on pretrained classifiers.
method Prepending a denoiser to any off-the-shelf classifier using randomized smoothing.
result Guaranteed ℓp-robustness to adversarial examples without modifying the pretrained classifier. Current neural network-based classifiers are susceptible to adversarial examples even in the black-box setting, where the attacker only has query access to the model. In practice, the threat model for real-world systems is often more restrictive than the typical black-box model where the adversary can observe the full …
In general, object identification tends not to work well on ambiguous, amorphous objects such as vegetation. In this study, we developed a simple but effective approach to identify ambiguous objects and applied the method to several moss species. As a result, the model correctly classified test images with accuracy mor…
Currently available methods for extracting saliency maps identify parts of the input which are the most important to a specific fixed classifier. We show that this strong dependence on a given classifier hinders their performance. To address this problem, we propose classifier-agnostic saliency map extraction, which fi…
The thesis introduces methods to use semantic hierarchy in image classification.
problem Limited work in training image classifiers with non-conventional external guidance.
method Injects label hierarchy knowledge into arbitrary classifiers and uses order-preserving embeddings for image classification.
result Both embedding-based models and CNN-classifiers with hierarchical information outperform a hierarchy-agnostic model.
TRAK traces model predictions to training data efficiently.
problem Inefficiency in data attribution methods for large-scale models.
method TRAK: a new data attribution method that is both effective and computationally tractable.
result TRAK matches the performance of methods requiring thousands of models with just a handful.
This paper analyzes β-Variational Classifiers for robustness and adversarial perturbation detection.
problem Limited robustness of deep neural networks in predictions.
method An analysis of β-Variational Classifiers, focusing on their robustness and adversarial perturbation detection. result Novel insights into the generative component of β-Variational Classifiers. New method reconstructs significant parts of training data from neural networks.
problem Understanding and reconstructing training data from neural networks.
method Proposes a novel reconstruction scheme based on recent theoretical results about neural network training.
result Shows that a significant fraction of training data can be reconstructed from neural network parameters.
In recent years there has been a sharp rise in networking applications, in which significant events need to be classified but only a few training instances are available. These are known as cases of one-shot learning. Examples include analyzing network traffic under zero-day attacks, and computer vision tasks by sensor…
This paper shows excessive invariance in adversarial robust models can make them more vulnerable to certain types of attacks.
problem Excessive invariance in adversarial robust models can make them more vulnerable to certain types of attacks.
method Analytical constructions and empirical studies of vision classifiers with state-of-the-art robustness to perturbation-based adversaries constrained by an ℓp norm. result Robustness to perturbation-based adversarial examples does not guarantee general robustness and can increase vulnerability to invariance-based adversarial examples.
We consider adversarial examples for image classification in the black-box decision-based setting. Here, an attacker cannot access confidence scores, but only the final label. Most attacks for this scenario are either unreliable or inefficient. Focusing on the latter, we show that a specific class of attacks, Boundary …
MSNet uses high frequency residual learning for efficient multi-scale image classification.
problem Efficient multi-scale image classification for mobile and embedded devices.
method Two network architecture: low resolution for low frequency, high resolution for high frequency residuals.
result MSNet achieves significant accuracy improvements over different base networks.
Machine learning models, especially neural network (NN) classifiers, are widely used in many applications including natural language processing, computer vision and cybersecurity. They provide high accuracy under the assumption of attack-free scenarios. However, this assumption has been defied by the introduction of ad…
WAR method improves classifier robustness in noisy label datasets.
problem Learning robust classifiers in presence of noisy labels.
method Adversarial regularization based on Wasserstein distance.
result WAR method outperforms state-of-the-art competitors on noisy label datasets.
The use of distributions and high-level features from deep architecture has become commonplace in modern computer vision. Both of these methodologies have separately achieved a great deal of success in many computer vision tasks. However, there has been little work attempting to leverage the power of these to methodolo…
Large-scale labeled training datasets have enabled deep neural networks to excel on a wide range of benchmark vision tasks. However, in many applications it is prohibitively expensive or time-consuming to obtain large quantities of labeled data. To cope with limited labeled training data, many have attempted to directl…
Sparse Vision MoE matches dense networks in image recognition while using less compute.
problem Scaling vision models efficiently in computer vision.
method Vision MoE (V-MoE) - a sparse version of Vision Transformer.
result V-MoE matches state-of-the-art dense networks in image recognition with half the compute.
Robust CLIP improves vision models' resistance to attacks.
problem Vulnerability of vision models to adversarial attacks.
method Unsupervised adversarial fine-tuning of CLIP.
result Robust CLIP prevents stealth attacks on vision models.
Classifiers based on sparse representations have recently been shown to provide excellent results in many visual recognition and classification tasks. However, the high cost of computing sparse representations at test time is a major obstacle that limits the applicability of these methods in large-scale problems, or in…
Deep neural networks outperform traditional methods in high-dimensional classification.
problem Understanding the empirical success of deep neural networks in high-dimensional classification.
method Proposed a teacher-student framework with Bayes classifier as ReLU neural networks, derived convergence rates for 0-1 and hinge losses.
result Sharp rate of convergence for classifiers trained using 0-1 or hinge loss, with Od(n−2/3) or Od(n−1) under separable data distribution. We propose a method to classify the causal relationship between two discrete variables given only the joint distribution of the variables, acknowledging that the method is subject to an inherent baseline error. We assume that the causal system is acyclicity, but we do allow for hidden common causes. Our algorithm presu…