Locally optimal detection of adversarial inputs to image classifiers

This research paper presents an advanced deep learning framework for tomato ripeness and maturity detection, employing Convolutional Neural Networks (CNNs) as the primary tool for automated classification. In particular, the study emphasizes the utilization of the ResNet-152 architecture, a sophisticated deep neural network known for its exceptional performance in image classification tasks. ResNet-152 addresses the challenges of fine-grained categorization by enabling the extraction of intricate visual features such as color nuances, textures, and shapes inherent to tomatoes. This eliminates the necessity for manual feature engineering, enhancing the model's ability to discern between "RIPE," "UNRIPE," and "ROTTEN" tomato classes with unprecedented accuracy. The ResNet-152 architecture's effectiveness lies in its unique design, featuring residual blocks that facilitate the training of very deep networks. This architecture mitigates the vanishing gradient problem and enables the model to learn complex hierarchical features, contributing to its state-of-the-art performance in image classification. In the context of tomato ripeness detection, ResNet-152 acts as a powerful tool for capturing and understanding the intricate visual cues that define different ripeness states, laying the foundation for an efficient and accurate automated sorting system in the agricultural and food processing industries.

Nowadays convolutional neural networks (CNNs) have produced the state-of-the-art performance in image classification and segmentation tasks. The efficiency of the neural networks is one of the bests when the testing samples are close to the training data. Nevertheless, if we make some transformation on the dataset, the performance of the convolutional neural network reduced. Recently, capsule networks (CapsNet) have been introduced to solve some of the problems of neural networks. In this paper we examine the effectiveness of three different capsule based neural networks, and compare the performance when the parameters of the dynamic routing algorithm and the squashing function are modified.

Medical image analysis using deep neural networks (DNN) has demonstrated state-of-the-art performance in image classification and segmentation tasks, aiding disease diagnosis. The accuracy of the DNN is largely governed by the quality and quantity of the data used to train the model. However, for the medical images, the critical security and privacy concerns regarding sharing of local medical data across medical establishments precludes exploiting the full DNN potential for clinical diagnosis. The federated learning (FL) approach enables the use of local model's parameters to train a global model, while ensuring data privacy and security. In this paper, we review the federated learning applications in medical image analysis with DNNs, highlight the security concerns, cover some efforts to improve FL model performance, and describe the challenges and future research directions.

Convolutional neural networks have demonstrated state-of-the-art performance in image classification and various other computer vision tasks. Plant disease detection is an important area of deep learning which has been addressed by many recent methods. However, there is a dire need to optimize these solutions for resource-constrained portable devices such as smartphones. This is a challenging problem because deep learning models are resource extensive in nature. This paper proposes an efficient method to systematically classify plant disease symptoms using convolutional neural networks. These networks are memory efficient and when coupled with the proposed training configuration it enables rapid development of industrial applications by reducing the training times. Another critical problem arises with the improper distribution of samples among classes known as the class imbalance problem, which is addressed by employing a simple statistical methodology. Transfer learning is a known technique for training small datasets which transfers pre-trained weights learned on a large dataset. However, during transfer learning, negative transfer learning is a common problem. Therefore, a stepwise transfer learning approach is proposed which can help in fast convergence, while reducing overfitting and preventing negative transfer learning during knowledge transfer across domains. The system is trained and evaluated on two plant disease datasets i.e., PlantVillage (a publicly available dataset) and pepper disease dataset provided by the National Institute of Horticultural and Herbal Science, Republic of Korea. The pepper dataset is particularly challenging as it contains images from different parts of the plant such as the leaf, pulp, and stem. The proposed system has dominated the previous works on the PlantVillage dataset and achieved 99% and 99.69% accuracy on the Pepper dataset and PlantVillage datasets, respectively.

Convolutional Neural Networks have demonstrated state-of-the-art performance in various image classification and computer vision-related tasks. Plant disease detection is one of the essential areas of image classification. Though many models have been proposed for the efficient classification of plant disease images, there is a dire need to develop an image classification mechanism based on deep learning, which has fewer parameters to implement the algorithm on mobile devices. In this manuscript, we propose a ConvPlant-Net, a Convolutional Neural Network based Plant disease detection system that uses a combination of Depth-Wise Separable Convolutional, 2-Dimensional transpose Layer, and a Convolutional layer for efficient learning of the high and low-level features. The proposed model contains only 31,998 trainable parameters. With fewer parameters, the model has achieved an accuracy of 98.62%, 99.36%, and 99.60% on Tomato, Pepper Bell, and Potato crops from the publicly available PlantVillage dataset.

This study introduces a framework aimed at bolstering the effectiveness of deep learning models in image classification tasks, with a specialized focus on diabetic retinopathy detection. The proposed strategy employs transfer and multitask learning methods for the consistent extraction of insights and refinement of pre-established models. The capability of multitask learning to simultaneously acquire knowledge from numerous related tasks, leveraging shared information and inter-dependencies, leads to considerable performance boosts. Integrated into this process is the feature guideline learning task, which empowers the model to have an attention to abnormal areas. This task is initially trained on a small collection of labeled images to develop a custom pre-trained model, which later forms the basis for subsequent classification tasks. The effectiveness of this proposed approach is validated through tests on various data sets. It has been proven that our method contributes considerable advancements in image classification performance. This research offers valuable perspectives on the detection and classification of diabetic retinopathy, providing solutions to problems originating from limited data, and shows promising potential for improving patient care.

Binary neural networks leverage Sign function to binarize weights and activations, which require gradient estimators to overcome its non-differentiability and will inevitably bring gradient errors during backpropagation. Although many hand-designed soft functions have been proposed as gradient estimators to better approximate gradients, their mechanism is not clear and there are still huge performance gaps between binary models and their full-precision counterparts. To address these issues and reduce gradient error, we propose to tackle network binarization as a binary classification problem and use a multi-layer perceptron (MLP) as the classifier in the forward pass and gradient estimator in the backward pass. Benefiting from the MLP’s theoretical capability to fit any continuous function, it can be adaptively learned to binarize networks and backpropagate gradients without any prior knowledge of soft functions. From this perspective, we further empirically justify that even a simple linear function can outperform previous complex soft functions. Extensive experiments demonstrate that the proposed method yields surprising performance both in image classification and human pose estimation tasks. Specifically, we achieve 65.7% top-1 accuracy of ResNet-34 on ImageNet dataset, with an absolute improvement of 2.6%. Moreover, we take binarization as a lightweighting approach for pose estimation models and propose well-designed binary pose estimation networks SBPN and BHRNet. When evaluating on the challenging Microsoft COCO keypoint dataset, the proposed method enables binary networks to achieve a mAP of up to 60.6 for the first time. Experiments conducted on real platforms demonstrate that BNN achieves a better balance between performance and computational complexity, especially when computational resources are extremely low.

Vision transformers have recently attained state-of-the-art results in visual recognition tasks. Their success is largely attributed to the self-attention component, which models the global dependencies among the image patches (tokens) and aggregates them into higher-level features. However, self-attention brings significant training difficulties to ViTs. Many recent works thus develop various new self-attention components to alleviate this issue. In this article, instead of developing complicated self-attention mechanism, we aim to explore simple approaches to fully release the potential of the vanilla self-attention. We first study the token selection behavior of self-attention and find that it suffers from a low diversity due to attention over-smoothing, which severely limits its effectiveness in learning discriminative token features. We then develop simple approaches to enhance selectivity and diversity for self-attention in token selection. The resulted token selector module can server as a drop-in module for various ViT backbones and consistently boost their performance. Significantly, they enable ViTs to achieve 84.6% top-1 classification accuracy on ImageNet with only 25M parameters. When scaled up to 81M parameters, the result can be further improved to 86.1%. In addition, we also present comprehensive experiments to demonstrate the token selector can be applied to a variety of transformer-based models to boost their performance for image classification, semantic segmentation and NLP tasks. Code is available at https://github.com/zhoudaquan/dvit_repo.

This study focuses on the optimization and practice of Convolutional Neural Networks (CNN) in image classification tasks. A deep neural network model was constructed by introducing the CBAM (Convolutional Block Attention Module) attention mechanism and residual structure. This model combines the stability of residual networks during training with the advantages of CBAM in feature extraction, aiming to improve image classification performance. The experiment used the CIFAR-10 dataset and achieved a testing accuracy of 86% through optimization methods such as data augmentation, batch normalization, reasonable control of the number of fully connected layers, and exponential decay learning rate strategy. The study also explored the effects of data augmentation, batch normalization layer utility, fully connected layer quantity control, residual structure effectiveness, and CBAM attention mechanism on model performance, providing valuable insights for optimizing CNN in image classification tasks.

Artificial neural networks based on sigmoidal neurons have achieved undisputed performance in various tasks, including image recognition and classification. The search for more biologically plausible artificial neuron models led to the creation of pulsed models. The new way of encoding information is through discrete pulses, and the timing of these pulses is important for the result. This work proposes the creation of a neural network with pulsed neurons for the task of image classification. The network uses cell models more similar to those found in animal brains, communicating through spikes and relying on a stochastic component for pulse generation. It also applies STDP as an unsupervised learning rule, very similar to human learning. Experiments were run using various parameter sets to study the network's dynamics in the image classification task. The results obtained were analyzed, and their performance indicates a promising method capable of good performance on three known image databases (Caltech 101, ETH-80, and MNIST). The database 1 achieved a classification accuracy of 87%, database 2 achieved 77% on ETH-80, and database 3 achieved 86% on MNIST respectively.

Locality-constrained affine subspace coding for image classification and retrieval

Chest X-ray analysis empowered with deep learning: A systematic review

In image classification tasks, commonly used deep learning network architectures include Convolutional Neural Network (CNN) and Vision Transformer (ViT). Still, both are relatively mature, while Graph Neural Network (GNN) network architecture has been attempted less in image classification. In addition, the architecture and design of GNN are still rapidly developing and have enormous potential. This study aims to create an image classification method based on superpixels and graph neural networks. This project attempted three different approaches to preserve superpixel node features with varying numbers of features, designed two training network architectures with different complexities and conducted six sets of tests on image classification tasks on the fashion-mnist dataset. As a result, the use of superpixel and GNN methods achieved good accuracy in image classification tasks, demonstrating the potential of this method in image recognition tasks. It was also found that adding the boundary coordinates of the region to the training in the selection of superpixel node features can improve the accuracy of the final image classification task.

Though traditional classification methods show well performance in classification tasks, most of them mainly lay emphasis on ‘classification’ rather than ‘cognition’. When a new object that has never been seen is encountered, the traditional methods falsely default the image as a certain category that has been studied, however, humans can first identify the image as new. In this paper, we present a memory model for visual images classification based on residual neural network and Bayesian decision (VICRB). First, the feature vectors of the visual images for each category are extracted with residual neural network and each feature component may be correctly copied or randomly produced. Then the processes about how the visual images represent, store and retrieve are modeled. The feature vector of test images is matched with the feature vectors of learned images and the likelihood ratio is computed according to probabilistic inference theory. Finally, the odds value in favor of an old over a new image is computed by all likelihood values. According to the odd value, the Bayesian decision rule is applied to the image classification. Experimental results on two benchmark images datasets show that the presented memory model performs well in images classification tasks.

To enhance the performance of image classification and speech recognition, the optimizer is considered an important factor for achieving high accuracy. The state-of-the-art optimizer can perform to serve in applications that may not require very high accuracy, yet the demand for high-precision image classification and speech recognition is increasing. This study implements an adaptive method for applying the particle filter technique with a gradient descent optimizer to improve model learning performance. Using a pretrained model helps reduce the computational time to deploy an image classification model and uses a simple deep convolutional neural network for speech recognition. The applied method results in a higher speech recognition accuracy score—89.693% for the test dataset—than the conventional method, which reaches 89.325%. The applied method also performs well on the image classification task, reaching an accuracy of 89.860% on the test dataset, better than the conventional method, which has an accuracy of 89.644%. Despite a slight difference in accuracy, the applied optimizer performs well in this dataset overall.