Using features from deep neural networks to model human categorization of natural images

Abstract Transfer learning from natural images is well used in deep neural networks (DNNs) for medical image classification to achieve computer-aided clinical diagnosis. Although the adversarial vulnerability of DNNs hinders practical applications owing to the high stakes of diagnosis, adversarial attacks are expected to be limited because training data — which are often required for adversarial attacks — are generally unavailable in terms of security and privacy preservation. Nevertheless, we hypothesized that adversarial attacks are also possible using natural images because pre-trained models do not change significantly after fine-tuning. We focused on three representative DNN-based medical image classification tasks (i.e., skin cancer, referable diabetic retinopathy, and pneumonia classifications) and investigated whether medical DNN models with transfer learning are vulnerable to universal adversarial perturbations (UAPs), generated using natural images. UAPs from natural images are useful for both non-targeted and targeted attacks. The performance of UAPs from natural images was significantly higher than that of random controls, although slightly lower than that of UAPs from training images. Vulnerability to UAPs from natural images was observed between different natural image datasets and between different model architectures. The use of transfer learning causes a security hole, which decreases the reliability and safety of computer-based disease diagnosis. Model training from random initialization (without transfer learning) reduced the performance of UAPs from natural images; however, it did not completely avoid vulnerability to UAPs. The vulnerability of UAPs from natural images will become a remarkable security threat.

Download Full-text

Deep neural networks for grape bunch segmentation in natural images from a consumer-grade camera

Precision Agriculture ◽

10.1007/s11119-020-09736-0 ◽

2020 ◽

Cited By ~ 1

Author(s):

R. Marani ◽

A. Milella ◽

A. Petitti ◽

G. Reina

Keyword(s):

Neural Networks ◽

Deep Neural Networks ◽

Natural Images

Download Full-text

Vision beyond Pixels: Visual Reasoning via Blocksworld Abstractions

Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence ◽

10.24963/ijcai.2019/907 ◽

2019 ◽

Author(s):

Tejas Gokhale

Keyword(s):

Neural Networks ◽

Computer Vision ◽

Cognitive Psychology ◽

Visual Perception ◽

Deep Neural Networks ◽

Natural Images ◽

Modular Approach ◽

Reasoning Task ◽

Two Stages ◽

Event Sequencing

Deep neural networks trained in an end-to-end fashion have brought about exceptional advances in computer vision, especially in computational perception. We go beyond perception and seek to enable vision modules to reason about perceived visual entities such as scenes, objects and actions. We introduce a challenging visual reasoning task, Image-Based Event Sequencing (IES) and compile the first IES dataset, Blocksworld Image Reasoning Dataset (BIRD). Motivated by the blocksworld concept, we propose a modular approach supported by literature in cognitive psychology and children's development. We decompose the problem into two stages - visual perception and event sequencing, and show that our approach can be extended to natural images without re-training.

Download Full-text

Artificial Intelligence Explained for Nonexperts

Seminars in Musculoskeletal Radiology ◽

10.1055/s-0039-3401041 ◽

2020 ◽

Vol 24 (01) ◽

pp. 003-011 ◽

Cited By ~ 1

Author(s):

Narges Razavian ◽

Florian Knoll ◽

Krzysztof J. Geras

Keyword(s):

Artificial Intelligence ◽

Neural Networks ◽

Clinical Practice ◽

Medical Imaging ◽

Deep Neural Networks ◽

Large Data ◽

Large Data Sets ◽

Natural Images ◽

Data Sets ◽

Current State

AbstractArtificial intelligence (AI) has made stunning progress in the last decade, made possible largely due to the advances in training deep neural networks with large data sets. Many of these solutions, initially developed for natural images, speech, or text, are now becoming successful in medical imaging. In this article we briefly summarize in an accessible way the current state of the field of AI. Furthermore, we highlight the most promising approaches and describe the current challenges that will need to be solved to enable broad deployment of AI in clinical practice.

Download Full-text

Controversial stimuli: Pitting neural networks against each other as models of human cognition

Proceedings of the National Academy of Sciences ◽

10.1073/pnas.1912334117 ◽

2020 ◽

Vol 117 (47) ◽

pp. 29330-29337 ◽

Cited By ~ 1

Author(s):

Tal Golan ◽

Prashant C. Raju ◽

Nikolaus Kriegeskorte

Keyword(s):

Neural Networks ◽

Visual Recognition ◽

Deep Neural Networks ◽

Human Subjects ◽

Human Perception ◽

Ground Truth ◽

Natural Images ◽

Human Cognition ◽

Distribution Models ◽

Adversarial Examples

Distinct scientific theories can make similar predictions. To adjudicate between theories, we must design experiments for which the theories make distinct predictions. Here we consider the problem of comparing deep neural networks as models of human visual recognition. To efficiently compare models’ ability to predict human responses, we synthesize controversial stimuli: images for which different models produce distinct responses. We applied this approach to two visual recognition tasks, handwritten digits (MNIST) and objects in small natural images (CIFAR-10). For each task, we synthesized controversial stimuli to maximize the disagreement among models which employed different architectures and recognition algorithms. Human subjects viewed hundreds of these stimuli, as well as natural examples, and judged the probability of presence of each digit/object category in each image. We quantified how accurately each model predicted the human judgments. The best-performing models were a generative analysis-by-synthesis model (based on variational autoencoders) for MNIST and a hybrid discriminative–generative joint energy model for CIFAR-10. These deep neural networks (DNNs), which model the distribution of images, performed better than purely discriminative DNNs, which learn only to map images to labels. None of the candidate models fully explained the human responses. Controversial stimuli generalize the concept of adversarial examples, obviating the need to assume a ground-truth model. Unlike natural images, controversial stimuli are not constrained to the stimulus distribution models are trained on, thus providing severe out-of-distribution tests that reveal the models’ inductive biases. Controversial stimuli therefore provide powerful probes of discrepancies between models and human perception.

Download Full-text

Optimized Deep Convolutional Neural Networks for Identification of Macular Diseases from Optical Coherence Tomography Images

Algorithms ◽

10.3390/a12030051 ◽

2019 ◽

Vol 12 (3) ◽

pp. 51 ◽

Cited By ~ 3

Author(s):

Qingge Ji ◽

Jie Huang ◽

Wenjie He ◽

Yankui Sun

Keyword(s):

Neural Networks ◽

Optical Coherence Tomography ◽

Large Scale ◽

Deep Neural Networks ◽

Natural Images ◽

Small Scale ◽

Optical Coherence ◽

Deep Convolutional Neural Networks ◽

Macular Diseases ◽

High Level

Finetuning pre-trained deep neural networks (DNN) delicately designed for large-scale natural images may not be suitable for medical images due to the intrinsic difference between the datasets. We propose a strategy to modify DNNs, which improves their performance on retinal optical coherence tomography (OCT) images. Deep features of pre-trained DNN are high-level features of natural images. These features harm the training of transfer learning. Our strategy is to remove some deep convolutional layers of the state-of-the-art pre-trained networks: GoogLeNet, ResNet and DenseNet. We try to find the optimized deep neural networks on small-scale and large-scale OCT datasets, respectively, in our experiments. Results show that optimized deep neural networks not only reduce computational burden, but also improve classification accuracy.

Download Full-text

Effect of levels in Dual Tree Complex Wavelet Transform when design Universal image stego-analytic

Iraqi Journal of Science ◽

10.24996/ijs.2020.61.3.23 ◽

2020 ◽

pp. 665-674

Author(s):

Hana Mohsen ◽

Halah Hasan

Keyword(s):

Neural Networks ◽

Wavelet Transform ◽

Deep Neural Networks ◽

Adaptive Methods ◽

Natural Images ◽

Detection Accuracy ◽

Complex Wavelet Transform ◽

Image Dataset ◽

Complex Wavelet ◽

Content Adaptive

Universal image stego-analytic has become an important issue due to the natural images features curse of dimensionality. Deep neural networks, especially deep convolution networks, have been widely used for the problem of universal image stegoanalytic design. This paper describes the effect of selecting suitable value for number of levels during image pre-processing with Dual Tree Complex Wavelet Transform. This value may significantly affect the detection accuracy which is obtained to evaluate the performance of the proposed system. The proposed system is evaluated using three content-adaptive methods, named Highly Undetetable steGO (HUGO), Wavelet Obtained Weights (WOW) and UNIversal WAvelet Relative Distortion (UNIWARD).The obtained precision 0.98387, 0.96659 and 0.98387 for the three content-adaptive methods, applied on BOSS image dataset, respectively. The obtained results show that number of level equals to 5 outperforms other numbers in terms of detection accuracy. Also it minimizes the ime required for both training and testing phases.

Download Full-text

Natural Images Allow Universal Adversarial Attacks on Medical Image Classification Using Deep Neural Networks With Transfer Learning

10.21203/rs.3.rs-757225/v2 ◽

2021 ◽

Author(s):

Akinori Minagi ◽

Hokuto Hirano ◽

Kazuhiro Takemoto

Keyword(s):

Neural Networks ◽

Image Classification ◽

Transfer Learning ◽

Medical Image ◽

Deep Neural Networks ◽

Disease Diagnosis ◽

Natural Images ◽

Fine Tuning ◽

Security And Privacy ◽

Medical Image Classification

Abstract Background. Transfer learning from natural images is well used in deep neural networks (DNNs) for medical image classification to achieve computer-aided clinical diagnosis. Although the adversarial vulnerability of DNNs hinders practical applications owing to the high stakes of diagnosis, adversarial attacks are expected to be limited because training data — which are often required for adversarial attacks — are generally unavailable in terms of security and privacy preservation. Nevertheless, we hypothesized that adversarial attacks are also possible using natural images because pre-trained models do not change significantly after fine-tuning.Methods. We considered three representative DNN-based medical image classification tasks (i.e., skin cancer, referable diabetic retinopathy, and pneumonia classifications) to investigate whether medical DNN models with transfer learning are vulnerable to universal adversarial perturbations (UAPs), generated using natural images.Results. UAPs from natural images are useful for both non-targeted and targeted attacks. The performance of UAPs from natural images was significantly higher than that of random controls, although slightly lower than that of UAPs from training images. Vulnerability to UAPs from natural images was observed between different natural image datasets and between different model architectures.Conclusion. The use of transfer learning causes a security hole, which decreases the reliability and safety of computer-based disease diagnosis. Model training from random initialization (without transfer learning) reduced the performance of UAPs from natural images; however, it did not completely avoid vulnerability to UAPs. The vulnerability of UAPs from natural images will become a remarkable security threat.

Download Full-text