Fashion MNIST Research Articles

Transfer learning-based Fashion Image Classification using Hybrid 2D-CNN and ImageNet Neural Network

Abstract: Many algorithms have been developed as a result of recent advances in machine learning to handle a variety of challenges. In recent years, the most popular transfer learning method has allowed researchers and engineers to run experiments with minimal computing and time resources. To tackle the challenges of classification, product identification, product suggestion, and picture-based search, this research proposed a transfer learning strategy for Fashion image classification based on hybrid 2D-CNN pretrained by VGG-16 and AlexNet. Pre-processing, feature extraction, and classification are the three parts of the proposed system's implementation. We used the Fashion MNIST dataset, which consists of 50,000 fashion photos that have been classified. Training and validation datasets have been separated. In comparison to other conventional methodologies, the suggested transfer learning approach has higher training and validation accuracy and reduced loss. Keywords: Machine Learning, Transfer Learning, Convolutional Neural Network, Image Classification, VGG16, AlexNet, 2D CNN.

Machine Learning Method for Functional Assessment of Retinal Models.

Challenges in the field of retinal prostheses motivate the development of retinal models to accurately simulate Retinal Ganglion Cells (RGCs) responses. The goal of retinal prostheses is to enable blind individuals to solve complex, reallife visual tasks. In this paper, we introduce the functional assessment (FA) of retinal models, which describes the concept of evaluating the performance of retinal models on visual understanding tasks. We present a machine learning method for FA: we feed traditional machine learning classifiers with RGC responses generated by retinal models, to solve object and digit recognition tasks (CIFAR-10, MNIST, Fashion MNIST, Imagenette). We examined critical FA aspects, including how the performance of FA depends on the task, how to optimally feed RGC responses to the classifiers and how the number of output neurons correlates with the model's accuracy. To increase the number of output neurons, we manipulated input images - by splitting and then feeding them to the retinal model and we found that image splitting does not significantly improve the model's accuracy. We also show that differences in the structure of datasets result in largely divergent performance of the retinal model (MNIST and Fashion MNIST exceeded 80% accuracy, while CIFAR-10 and Imagenette achieved ∼40%). Furthermore, retinal models which perform better in standard evaluation, i.e. more accurately predict RGC response, perform better in FA as well. However, unlike standard evaluation, FA results can be straightforwardly interpreted in the context of comparing the quality of visual perception.

Incoherent reconstruction-free object recognition with mask-based lensless optics and the Transformer.

A mask-based lensless camera adopts a thin mask to optically encode the scene and records the encoded pattern on an image sensor. The lensless camera can be thinner, lighter and cheaper than the lensed camera. But additional computation is required to reconstruct an image from the encoded pattern. Considering that the significant application of the lensless camera could be inference, we propose to perform object recognition directly on the encoded pattern. Avoiding image reconstruction not only saves computational resources but also averts errors and artifacts in reconstruction. We theoretically analyze multiplexing property in mask-based lensless optics which maps local information in the scene to overlapping global information in the encoded pattern. To better extract global features, we propose a simplified Transformer-based architecture. This is the first time to study Transformer-based architecture for encoded pattern recognition in mask-based lensless optics. In the optical experiment, the proposed system achieves 91.47% accuracy on the Fashion MNIST and 96.64% ROC AUC on the cats-vs-dogs dataset. The feasibility of physical object recognition is also evaluated.

Entangled q-convolutional neural nets

We introduce a machine learning model, the q-CNN model, sharing key features with convolutional neural networks and admitting a tensor network description. As examples, we apply q-CNN to the MNIST and Fashion MNIST classification tasks. We explain how the network associates a quantum state to each classification label, and study the entanglement structure of these network states. In both our experiments on the MNIST and Fashion-MNIST datasets, we observe a distinct increase in both the left/right as well as the up/down bipartition entanglement entropy (EE) during training as the network learns the fine features of the data. More generally, we observe a universal negative correlation between the value of the EE and the value of the cost function, suggesting that the network needs to learn the entanglement structure in order the perform the task accurately. This supports the possibility of exploiting the entanglement structure as a guide to design the machine learning algorithm suitable for given tasks.

Reliability Enhancement of Inverter-Based Memristor Crossbar Neural Networks Using Mathematical Analysis of Circuit Non-Idealities

In this paper, the sensitivity of the neural network (NN) outputs to device parameter uncertainties (non-idealities) in inverter-based memristor ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IM</i> ) crossbar neuromorphic circuits is mathematically modeled and verified using exhaustive circuit and system-level simulations. The NN sensitivity is obtained by modeling the sensitivity of the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IM</i> neuron output to the non-idealities of its circuit elements. The analysis reveals a higher sensitivity of the output voltage of the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IM</i> neuron to the non-idealities of the inverters compared to the conductance variation of the memristors. Among the inverter non-idealities, horizontal shift of the inverters voltage transfer characteristic (VTC) shows the highest impact on the output voltage of the neuron. To reduce the accuracy loss due to the variations, a training approach which includes a sensitivity term in the cost function of the training phase, is suggested. The achievable improvements through the said NN training approach are evaluated. In the evaluation, the California Housing, MNIST, and Fashion MNIST datasets are employed. The results show up to 50% reduction in the NN output variations in the presence of circuit elements’ non-idealities.

Distinguishing Lightweight Block Ciphers in Encrypted Images

Modern day lightweight block ciphers provide powerful encryption methods for securing IoT communication data. Tiny digital devices exchange private data which the individual users might not be willing to get disclosed. On the other hand, the adversaries try their level best to capture this private data. The first step towards this is to identify the encryption scheme. This work is an effort to construct a distinguisher to identify the cipher used in encrypting the traffic data. We try to establish a deep learning based method to identify the encryption scheme used from a set of three lightweight block ciphers viz. LBlock, PRESENT and SPECK. We make use of images from MNIST and fashion MNIST data sets for establishing the cryptographic distinguisher. Our results show that the overall classification accuracy depends firstly on the type of key used in encryption and secondly on how frequently the pixel values change in original input image.

Research on Image Classification Algorithm Based on Pytorch

With the development of artificial intelligence and deep learning, image classification technology has ushered in new opportunities and challenges. The so-called image classification problem is the problem that the user passes in the image, and then the computer sends out the classification and description of the content of the image. Feature description and detection as a traditional image classification method have many drawbacks, such as low accuracy and long time-consuming problems. Among the existing deep learning frameworks, Pytorch is relatively effective. Researchers can use the Pytorch framework to construct convolutional neural network models quickly and easily, and train them on massive data sets. This paper briefly introduces the image classification algorithm, neural network and Pytorch. The classification experiment is carried out on the Fashion MNIST data set, and the accuracy of the model reaches 88.39%. The future research direction is discussed.

DQRE-SCnet: A novel hybrid approach for selecting users in Federated Learning with Deep-Q-Reinforcement Learning based on Spectral Clustering

Machine learning models based on sensitive data in the real-world promise advances in areas ranging from medical screening to disease outbreaks, agriculture, industry, defense science, and more. In many applications, learning participant communication rounds benefit from collecting their own private data sets, teaching detailed machine learning models on the real data, and sharing the benefits of using these models. Due to existing privacy and security concerns, most people avoid sensitive data sharing for training. Without each user demonstrating their local data to a central server, Federated Learning allows various parties to train a machine learning algorithm on their shared data jointly. This method of collective privacy learning results in the expense of important communication during training. Most large-scale machine learning applications require decentralized learning based on data sets generated on various devices and places. Such datasets represent an essential obstacle to decentralized learning, as their diverse contexts contribute to significant differences in the delivery of data across devices and locations. Researchers have proposed several ways to achieve data privacy in Federated Learning systems. However, there are still challenges with homogeneous local data. This research’s approach is to select nodes (users) to share their data in Federated Learning for independent data-based equilibrium to improve accuracy, reduce training time, and increase convergence. Therefore, this research presents a combined Deep-Q-Reinforcement Learning Ensemble based on Spectral Clustering called DQRE-SCnet to choose a subset of devices in each communication round. Based on the results, it has been displayed that it is possible to decrease the number of communication rounds needed in Federated Learning. The realized reduction in the communication rounds are 51%, 25%, and 44% on the three datasets MNIST, Fashion MNIST, and CIFAR-10, respectively.

Synaptic metaplasticity for image processing enhancement in convolutional neural networks

Synaptic metaplasticity is a biological phenomenon shortly defined as the plasticity of synaptic plasticity, meaning that the previous history of the synaptic activity determines its current plasticity. This phenomenon interferes with some of the underlying mechanisms that are considered important in memory and learning processes, such as long-term potentiation and long-term depression. In this work, we provide an approach to include metaplasticity in convolutional neural networks to enhance learning in image classification problems. This approach consists of including metaplasticity as a weight update function in the backpropagation stage of convolutional layers. To validate this proposal, we have been used eight different award-winning convolutional neural networks architectures: LeNet-5, AlexNet, GoogLeNet, VGG16, VGG32, ResNet50, DenseNet121 and DenseNet169; trained with four different popular datasets for benchmarking: MNIST, Fashion MNIST, CIFAR-10 and CIFAR-100. Experimental results show that there is a performance enhancement for each of the convolution neural network architectures in all the datasets used.

Progressive Convolutional Neural Network for Incremental Learning

In this paper, we present a novel incremental learning technique to solve the catastrophic forgetting problem observed in the CNN architectures. We used a progressive deep neural network to incrementally learn new classes while keeping the performance of the network unchanged on old classes. The incremental training requires us to train the network only for new classes and fine-tune the final fully connected layer, without needing to train the entire network again, which significantly reduces the training time. We evaluate the proposed architecture extensively on image classification task using Fashion MNIST, CIFAR-100 and ImageNet-1000 datasets. Experimental results show that the proposed network architecture not only alleviates catastrophic forgetting but can also leverages prior knowledge via lateral connections to previously learned classes and their features. In addition, the proposed scheme is easily scalable and does not require structural changes on the network trained on the old task, which are highly required properties in embedded systems.

Loading-Aware Reliability Improvement of Ultra-Low Power Memristive Neural Networks

In this paper, a method for offline training of inverter-based memristive neural networks (IM-NNs), called ERIM, is presented. In this method, the output voltage of the inverter is modeled very accurately by considering the loading effect of the memristive crossbar. To properly choose the size of each inverter, its output load and the required slope of its voltage transfer characteristic (VTC) for an acceptable level of resiliency to the circuit element non-idealities are taken into account. The efficacy of ERIM is investigated by comparing its accuracy to those of two recently proposed offline training methods for IM-NNs (RIM and PHAX). The study is performed using IRIS, BCW, MNIST, and Fashion MNIST datasets. Simulation results show that 72% (56%) reduction in average energy consumption of the trained networks is achieved compared to RIM (PHAX) thanks to proper sizing of the inverters. In addition, due to the higher accuracy of the NN mathematical model, ERIM results in significant improvements in the match between the results of high-level modeling and HSPICE simulations while exhibiting lower sensitivity to circuit element variations.

Conditional information gain networks as sparse mixture of experts

Deep neural network models owe their representational power and high performance in classification tasks to the high number of learnable parameters. Running deep neural network models in limited-resource environments is a problematic task. Models employing conditional computing aim to reduce the computational burden while retaining model performance on par with more complex neural network models. This paper, proposes a new model, Conditional Information Gain Networks as Sparse Mixture of Experts (sMoE-CIGNs). A CIGN model is a neural tree that allows conditionally skipping parts of the tree based on routing mechanisms inserted into the architecture. These routing mechanisms are based on differentiable Information Gain objectives. CIGN groups semantically similar samples in the leaves, enabling simpler classifiers to focus on differentiating between similar classes. This lets the CIGN model attain high classification performances with lighter models. We further improve the basic CIGN model by proposing a sparse mixture of experts model for difficult to classify samples that may get routed to suboptimal branches. If a sample has routing confidence higher than a specific threshold, the sample may be routed towards multiple child nodes. The classification decision can then be taken as a mixture of these expert decisions. We learn the optimal routing thresholds by Bayesian Optimization over a validation set by minimizing a weighted loss, including the classification accuracy and the number of multiplication and accumulations (MAC). We show the effectiveness of the CIGN models enhanced with the Sparse Mixture of Experts approach with extensive tests on MNIST, Fashion MNIST, CIFAR 100 and UCI-USPS datasets, as well as comparisons with methods from the literature. sMoE-CIGN models can retain high generalization performance, on par with a thick unconditional model while keeping the operation burden at the same level with a much thinner model.11This article is an extended version of reference “Bicici U.C., Keskin C., Akarun L., Conditional information gain networks”, which appeared in Proceedings of the 24th International Conference on Pattern Recognition (ICPR) (2018).

Speed Reading in the Dark: Accelerating Functional Encryption for Quadratic Functions with Reprogrammable Hardware

Functional encryption is a new paradigm for encryption where decryption does not give the entire plaintext but only some function of it. Functional encryption has great potential in privacy-enhancing technologies but suffers from excessive computational overheads. We introduce the first hardware accelerator that supports functional encryption for quadratic functions. Our accelerator is implemented on a reprogrammable system-on-chip following the hardware/software codesign methogology. We benchmark our implementation for two privacy-preserving machine learning applications: (1) classification of handwritten digits from the MNIST database and (2) classification of clothes images from the Fashion MNIST database. In both cases, classification is performed with encrypted images. We show that our implementation offers speedups of over 200 times compared to a published software implementation and permits applications which are unfeasible with software-only solutions.

Dynamically Optimizing Network Structure Based on Synaptic Pruning in the Brain.

Most neural networks need to predefine the network architecture empirically, which may cause over-fitting or under-fitting. Besides, a large number of parameters in a fully connected network leads to the prohibitively expensive computational cost and storage overhead, which makes the model hard to be deployed on mobile devices. Dynamically optimizing the network architecture by pruning unused synapses is a promising technique for solving this problem. Most existing pruning methods focus on reducing the redundancy of deep convolutional neural networks by pruning unimportant filters or weights, at the cost of accuracy drop. In this paper, we propose an effective brain-inspired synaptic pruning method to dynamically modulate the network architecture and simultaneously improve network performance. The proposed model is biologically inspired as it dynamically eliminates redundant connections based on the synaptic pruning rules used during the brain's development. Connections are pruned if they are not activated or less activated multiple times consecutively. Extensive experiments demonstrate the effectiveness of our method on classification tasks of different complexity with the MNIST, Fashion MNIST, and CIFAR-10 datasets. Experimental results reveal that even for a compact network, the proposed method can also remove up to 59–90% of the connections, with relative improvement in learning speed and accuracy.

Differentially private GANs by adding noise to Discriminator’s loss

Differentially private generative adversary network (GAN) is a very promising field in data privacy, with many practical real-world applications. The idea of differentially private GAN is to provide differential privacy protection for sensitive training datasets. By using differentially private GANs, training datasets can be protected from being remembered or encoded into GAN’s parameters or generated data. Therefore, generated data can be safely used for data augmentation or replacing real sensitive data with very little privacy loss. However, existing methods for differentially private GANs are notoriously inefficient, most of them use a modified Tensorflow library, i.e., Tensorflow Privacy provided by Google, which is not only hard for programming but also very time-consuming even on TPU. In this paper, we provide a simpler and more efficient way to achieve differentially private GANs: when we set the training process of a GAN under a certain manner, we can use discriminator’s loss as vehicle to achieve differential privacy. Compared with existing methods, our method needs no modification on Tensorflow core functions and will not significantly drag down the training process. We test our method on GANs for real-world datasets MNIST, Fashion MNIST, and SVHN datasets, experimental results show that our method is easy to implement, more practical, and more efficient than existing methods. With these advantages, we believe our method for differentially private GANs will be widely used in the very near feature.

Study of Image Classification Accuracy with Fourier Ptychography

In this research, the accuracy of image classification with Fourier Ptychography Microscopy (FPM) has been systematically investigated. Multiple linear regression shows a strong linear relationship between the results of image classification accuracy and image visual appearance quality based on PSNR and SSIM with multiple training datasets including MINST, Fashion MNIST, Cifar, Caltech 101, and customized training datasets. It is, therefore, feasible to predict the image classification accuracy only based on PSNR and SSIM. It is also found that the image classification accuracy of FPM reconstructed with higher resolution images is significantly different from using the lower resolution images under the lower numerical aperture (NA) condition. The difference is yet less pronounced under the higher NA condition.

Might a Single Neuron Solve Interesting Machine Learning Problems Through Successive Computations on Its Dendritic Tree?

Physiological experiments have highlighted how the dendrites of biological neurons can nonlinearly process distributed synaptic inputs. However, it is unclear how aspects of a dendritic tree, such as its branched morphology or its repetition of presynaptic inputs, determine neural computation beyond this apparent nonlinearity. Here we use a simple model where the dendrite is implemented as a sequence of thresholded linear units. We manipulate the architecture of this model to investigate the impacts of binary branching constraints and repetition of synaptic inputs on neural computation. We find that models with such manipulations can perform well on machine learning tasks, such as Fashion MNIST or Extended MNIST. We find that model performance on these tasks is limited by binary tree branching and dendritic asymmetry and is improved by the repetition of synaptic inputs to different dendritic branches. These computational experiments further neuroscience theory on how different dendritic properties might determine neural computation of clearly defined tasks.

Image Enhanced Machine Vision System for Smart Factory

Machine vision is a technology that helps the computer as if a person recognizes and determines things. In recent years, as advanced technologies such as optical systems, artificial intelligence and big data advanced in conventional machine vision system became more accurate quality inspection and it increases the manufacturing efficiency. In machine vision systems using deep learning, the image quality of the input image is very important. However, most images obtained in the industrial field for quality inspection typically contain noise. This noise is a major factor in the performance of the machine vision system. Therefore, in order to improve the performance of the machine vision system, it is necessary to eliminate the noise of the image. There are lots of research being done to remove noise from the image. In this paper, we propose an autoencoder based machine vision system to eliminate noise in the image. Through experiment proposed model showed better performance compared to the basic autoencoder model in denoising and image reconstruction capability for MNIST and fashion MNIST data sets.

Multivariate‐bounded Gaussian mixture model with minimum message length criterion for model selection

AbstractBounded support Gaussian mixture model (BGMM) has been proposed for data modelling as an alternative to unbounded support mixture models for the cases when the data lies in bounded support. In this paper, we propose applications of multivariate BGMM in data clustering for more insightful analysis of the model. We also propose minimum message length (MML) criterion for model selection in data clustering using multivariate BGMM. The presented model is applied to data clustering in several speech (TSP and Spoken Digits) and image databases (MNIST and Fashion MNIST). We also propose the application of BGMM in code‐book generation at feature extraction phase. Inspired by the success of bag of visual words approach in computer vision, it is also introduced in speech data representation and validated through experiments presented in this paper. For validation of model selection criterion, MML is applied to different medical, speech and image datasets. Experimental results obtained during the model selection through MML are further compared with seven different model selection criteria. The results presented in the paper demonstrate the effectiveness of BGMM for clustering speech and image databases, code‐book generation through clustering for feature representation and model selection.

Deep Reinforcement Learning Assisted Federated Learning Algorithm for Data Management of IIoT

The continuous expanded scale of the industrial Internet of Things (IIoT) leads to IIoT equipments generating massive amounts of user data every moment. According to the different requirement of end users, these data usually have high heterogeneity and privacy, while most of users are reluctant to expose them to the public view. How to manage these time series data in an efficient and safe way in the field of IIoT is still an open issue, such that it has attracted extensive attention from academia and industry. As a new machine learning paradigm, federated learning (FL) has great advantages in training heterogeneous and private data. This article studies the FL technology applications to manage IIoT equipment data in wireless network environments. In order to increase the model aggregation rate and reduce communication costs, we apply deep reinforcement learning (DRL) to IIoT equipment selection process, specifically to select those IIoT equipment nodes with accurate models. Therefore, we propose a FL algorithm assisted by DRL, which can take into account the privacy and efficiency of data training of IIoT equipment. By analyzing the data characteristics of IIoT equipments, we use MNIST, fashion MNIST, and CIFAR-10 datasets to represent the data generated by IIoT. During the experiment, we employ the deep neural network model to train the data, and experimental results show that the accuracy can reach more than 97%, which corroborates the effectiveness of the proposed algorithm.