Caffe Deep Learning Framework Research Articles

Ultra low-power, wearable, accelerated shallow-learning fall detection for elderly at-risk persons

This work focuses on the development and manufacturing of a wireless, wearable, low-power fall detection sensor (FDS) designed to predict and detect falls in elderly at-risk individuals. Unintentional falls are a significant risk in this demographic, often resulting from diminished physical capabilities such as reduced hand grip strength and complications from conditions like arthritis, vertigo, and neuromuscular issues. To address this, we utilize advanced low-power field-programmable gate arrays (FPGAs) to implement a fixed-function neural network capable of categorizing activities of daily life (ADLs), including the detection of falls. This system employs a Convolutional Neural Network (CNN) model, trained and validated using the Caffe deep learning framework with data collected from human subjects experiments. This system integrates an ST Microelectronics LSM6DSOX inertial measurement unit (IMU) sensor, embedded with an ultra-low-power Lattice iCE40UP FPGA, which samples and stores joint acceleration and orientation rate. Additionally, we have acquired and published a dataset of 3D accelerometer and gyroscope measurements from predefined ADLs and falls, using volunteer human subjects. This innovative approach aims to enhance the safety and well-being of older adults by providing timely and accurate fall detection and prediction.In this paper, we present an innovative approach to utilizing a compact Convolutional Neural Network (CNN) core for accelerating convolutional operations on a machine learning model, suitable for deployment on an ultra-low power FPGA.

Cross-Model Hashing Retrieval Based on Deep Residual Network

In the era of big data rich in We Media, the single mode retrieval system has been unable to meet people’s demand for information retrieval. This paper proposes a new solution to the problem of feature extraction and unified mapping of different modes: A Cross-Modal Hashing retrieval algorithm based on Deep Residual Network (CMHR-DRN). The model construction is divided into two stages: The first stage is the feature extraction of different modal data, including the use of Deep Residual Network (DRN) to extract the image features, using the method of combining TF-IDF with the full connection network to extract the text features, and the obtained image and text features used as the input of the second stage. In the second stage, the image and text features are mapped into Hash functions by supervised learning, and the image and text features are mapped to the common binary Hamming space. In the process of mapping, the distance measurement of the original distance measurement and the common feature space are kept unchanged as far as possible to improve the accuracy of Cross-Modal Retrieval. In training the model, adaptive moment estimation (Adam) is used to calculate the adaptive learning rate of each parameter, and the stochastic gradient descent (SGD) is calculated to obtain the minimum loss function. The whole training process is completed on Caffe deep learning framework. Experiments show that the proposed algorithm CMHR-DRN based on Deep Residual Network has better retrieval performance and stronger advantages than other Cross-Modal algorithms CMFH, CMDN and CMSSH.

A deep learning network-assisted bladder tumour recognition under cystoscopy based on Caffe deep learning framework and EasyDL platform.

Cystoscopy plays an important role in the diagnosis of bladder tumours. As a typical representative of the deep learning algorithm, the convolutional neural network has shown great advantages in the field of image recognition and segmentation. One thousand two photographs of normal bladder tissue and 734 photos of bladder tumours under cystoscopy were taken from 175 patients. Caffe deep learning framework and EasyDL platform were used to structure and train the model. The trained model from the EasyDL platform was deployed on a mobile phone. The accuracy rate of the neural network to recognise the bladder cancer based on Caffe framework was 82.9%, and the data on the EasyDL platform were 96.9%. The model from EasyDL platform could discern bladder cancer accurately on the phone and website. The deep learning network could recognise the bladder cancer accurately. Deploying that model on the mobile phone was useful for clinical use.

Image Classification Search System based on Deep Learning Method

Image classification is to distinguish different types of images based on image information. It is an important basic issue in computer vision, and is also the fundamental for image detection, image segmentation, object tracking, and behavior analysis. Deep learning is a new field in machine learning research. Its motivation is to simulate the neural network of the human brain for analytical learning. Like the human brain, deep learning can interpret the data of images, sounds, and texts. The system is based on the Caffe deep learning framework. Firstly, the data set is trained and analyzed, and a model based on deep learning network is built to obtain the image feature information and corresponding data classification. Then the target image is expanded based on the bvlc-imagenet training set model, and finally achieve "search an image with an image" web application.

D y VED eep

Deep Neural Networks (DNNs) have advanced the state-of-the-art in a variety of machine learning tasks and are deployed in increasing numbers of products and services. However, the computational requirements of training and evaluating large-scale DNNs are growing at a much faster pace than the capabilities of the underlying hardware platforms that they are executed upon. To address this challenge, one promising approach is to exploit the error resilient nature of DNNs by skipping or approximating computations that have negligible impact on classification accuracy. Almost all prior efforts in this direction propose static DNN approximations by either pruning network connections, implementing computations at lower precision, or compressing weights. In this work, we propose <u>Dy</u>namic <u>V</u>ariable <u>E</u>ffort <u>Deep</u> Neural Networks (D y VED eep ) to reduce the computational requirements of DNNs during inference. Complementary to the aforementioned static approaches, DyVEDeep is a dynamic approach that exploits heterogeneity in the DNN inputs to improve their compute efficiency with comparable classification accuracy and without requiring any re-training. D y VED eep equips DNNs with dynamic effort mechanisms that identify computations critical to classifying a given input and focus computational effort only on the critical computations, while skipping or approximating the rest. We propose three dynamic effort mechanisms that operate at different levels of granularity viz. neuron, feature, and layer levels. We build D y VED eep versions of six popular image recognition benchmarks (CIFAR-10, AlexNet, OverFeat, VGG-16, SqueezeNet, and Deep-Compressed-AlexNet) within the Caffe deep-learning framework. We evaluate D y VED eep on two platforms—a high-performance server with a 2.7 GHz Intel Xeon E5-2680 processor and 128 GB memory, and a low-power Raspberry Pi board with an ARM Cortex A53 processor and 1 GB memory. Across all benchmarks, D y VED eep achieves 2.47×--5.15× reduction in the number of scalar operations, which translates to 1.94×--2.23× and 1.46×--3.46× performance improvement over well-optimized baselines on the Xeon server and the Raspberry Pi, respectively, with comparable classification accuracy.

Caffeine: Toward Uniformed Representation and Acceleration for Deep Convolutional Neural Networks

With the recent advancement of multilayer convolutional neural networks (CNNs) and fully connected networks (FCNs), deep learning has achieved amazing success in many areas, especially in visual content understanding and classification. To improve the performance and energy efficiency of the computation-demanding CNN, the FPGA-based acceleration emerges as one of the most attractive alternatives. In this paper, we design and implement Caffeine, a hardware/software co-designed library to efficiently accelerate the entire CNN and FCN on FPGAs. First, we propose a uniformed convolutional matrix-multiplication representation for both computation-bound convolutional layers and communication-bound FCN layers. Based on this representation, we optimize the accelerator micro-architecture and maximize the underlying FPGA computing and bandwidth resource utilization based on a revised roofline model. Moreover, we design an automation flow to directly compile high-level network definitions to the final FPGA accelerator. As a case study, we integrate Caffeine into the industry-standard software deep learning framework Caffe. We evaluate Caffeine and its integration with Caffe by implementing VGG16 and AlexNet networks on multiple FPGA platforms. Caffeine achieves a peak performance of 1460 giga fixed point operations per second on a medium-sized Xilinx KU060 FPGA board; to our knowledge, this is the best published result. It achieves more than $100\boldsymbol {\times }$ speed-up on FCN layers over prior FPGA accelerators. An end-to-end evaluation with Caffe integration shows up to $29\boldsymbol {\times }$ and $150\boldsymbol {\times }$ performance and energy gains over Caffe on a 12-core Xeon server, and $5.7\boldsymbol {\times }$ better energy efficiency over the GPU implementation. Performance projections for a system with a high-end FPGA (Virtex7 690t) show even higher gains.

SparCE: Sparsity Aware General-Purpose Core Extensions to Accelerate Deep Neural Networks

Deep Neural Networks (DNNs) have emerged as the method of choice for solving a wide range of machine learning tasks. The enormous computational demand posed by DNNs is a key challenge for computing system designers and has most commonly been addressed through the design of DNN accelerators. However, these specialized accelerators utilize large quantities of multiply-accumulate units and on-chip memory and are prohibitive in area and cost constrained systems such as wearable devices and IoT sensors. In this work, we take a complementary approach and improve the performance of DNNs on general-purpose processor (GPP) cores. We do so by exploiting a key attribute of DNNs, viz. sparsity or the prevalence of zero values. We propose Sparsity-aware Core Extensions (SparCE) - a set of low-overhead micro-architectural and ISA extensions that dynamically detect whether an operand (e.g., the result of a load instruction) is zero and subsequently skip a set of future instructions that use it. To maximize performance benefits, SparCE ensures that the instructions to be skipped are prevented from even being fetched, as squashing instructions comes with a penalty (e.g., a pipeline stall). SparCE consists of 2 key micro-architectural enhancements. First, a Sparsity Register File (SpRF) is utilized to track registers that are zero. Next, a Sparsity-Aware Skip Address (SASA) Table is used to indicate instruction sequences that can be skipped, and to specify conditions on SpRF registers that trigger instruction skipping. When an instruction is fetched, SparCE dynamically pre-identifies whether the following instruction(s) can be skipped, and if so appropriately modifies the program counter, thereby skipping the redundant instructions and improving performance. We model SparCE using the gem5 architectural simulator, and evaluate our approach on 6 state-of-the-art image-recognition DNNs in the context of both training and inference using the Caffe deep learning framework. On a scalar microprocessor, SparCE achieves 1.11×-1.96× speedups across both convolution and fully-connected layers that exhibit 10-90 percent sparsity. These speedups translate to 19-31 percent reduction in execution time at the overall application-level. We also evaluate SparCE on a 4-way SIMD ARMv8 processor using the OpenBLAS library, and demonstrate that SparCE achieves 8-15 percent reduction in the application-level execution time.

A DEEP CONVOLUTIONAL AUTO-ENCODER WITH POOLING – UNPOOLING LAYERS IN CAFFE

This paper presents the development of several models of a deep convolutional auto-encoder in the Caffe deep learning framework and their experimental evaluation on the example of MNIST dataset. We have created five models of a convolutional auto-encoder which differ architecturally by the presence or absence of pooling and unpooling layers in the auto-encoder’s encoder and decoder parts. Our results show that the developed models provide very good results in dimensionality reduction and unsupervised clustering tasks, and small classification errors when we used the learned internal code as an input of a supervised linear classifier and multi-layer perceptron. The best results were provided by a model where the encoder part contains convolutional and pooling layers, followed by an analogous decoder part with deconvolution and unpooling layers without the use of switch variables in the decoder part. The paper also discusses practical details of the creation of a deep convolutional auto-encoder in the very popular Caffe deep learning framework. We believe that our approach and results presented in this paper could help other researchers to build efficient deep neural network architectures in the future.

Exploiting Parallelism for CNN Applications on 3D Stacked Processing-In-Memory Architecture

Deep convolutional neural networks (CNNs) are widely adopted in intelligent systems with unprecedented accuracy but at the cost of a substantial amount of data movement. Although the emerging processing-in-memory (PIM) architecture seeks to minimize data movement by placing memory near processing elements, memory is still the major bottleneck in the entire system. The selection of hyper-parameters in the training of CNN applications requires over hundreds of kilobytes cache capacity for concurrent processing of convolutions. How to jointly explore the computation capability of the PIM architecture and the highly parallel property of neural networks remains a critical issue. This paper presents Para-Net , that exploits Para llelism for deterministic convolutional neural Net works on the PIM architecture. Para- Net achieves data-level parallelism for convolutions by fully utilizing the on-chip processing engine (PE) in PIM. The objective is to capture the characteristics of neural networks and present a hardware-independent design to jointly optimize the scheduling of both intermediate results and computation tasks. We formulate this data allocation problem as a dynamic programming model and obtain an optimal solution. To demonstrate the viability of the proposed Para-Net, we conduct a set of experiments using a variety of realistic CNN applications. The graph abstractions are obtained from deep learning framework Caffe. Experimental results show that Para-Net can significantly reduce processing time and improve cache efficiency compared to representative schemes.

Research of animals image semantic segmentation based on deep learning

SummaryIt is imperative for us to develop the technology of image semantic segmentation with the increasing demand in the image processing. Nowadays, the development of deep learning is of great significance to the improvement of image segmentation. Furthermore, the paper discussed the relationship between image semantic segmentation and animal image research based on the actual situation, and found that animal image processing technology plays a more important role in the field of protecting precious animals. The end‐to‐end network training of this paper is consisted of Fully Convolutional Network (FCN) for the front end and Conditional Random Fields as Recurrent Neural Networks (CRF‐RNN) for the back end via comparing a variety of research methods. The experiments achieved desired outcome for the semantic segmentation of animal images by utilizing Caffe deep learning framework and explained the implementation details from the aspects of training and testing.

Myocardial segmentation in cardiac magnetic resonance images using fully convolutional neural networks

Abstract According to the World Health Organization, cardiovascular diseases are the leading cause of death worldwide. Many coronary diseases involve the left ventricle; therefore, estimation of several functional parameters from a previous segmentation of this structure can be helpful in diagnosis. Although a high number of automated methods have been proposed, left ventricle segmentation in cardiac MRI images remains an open problem. In this work we propose a deep fully convolutional neural network architecture to address this issue and assess its performance. The model was trained end to end in a supervised learning stage from whole image input and ground truths to make a per pixel classification in order to segment the myocardium. For its design, development and experimentation a Caffe deep learning framework over an NVidia Quadro K4200 Graphics Processing Unit was used. Training and testing processes were carried out using 10-fold cross validation with short axis images. In addition, the performance of six optimization methods was compared. The proposed model was validated in 45 datasets of Sunnybrook database using a Dice coefficient, Average Perpendicular Distance (APD) and percentage of good contours (GC) metrics and compared with other state-of-the-art approaches. Results show the robustness and feasibility of the proposed method.

Human gait recognition based on Caffe deep learning framework

Human gait recognition based on Caffe deep learning framework

Performance Evaluation of Pipeline-Based Processing for the Caffe Deep Learning Framework

Performance Evaluation of Pipeline-Based Processing for the Caffe Deep Learning Framework

Deep-learning: investigating deep neural networks hyper-parameters and comparison of performance to shallow methods for modeling bioactivity data

BackgroundIn recent years, research in artificial neural networks has resurged, now under the deep-learning umbrella, and grown extremely popular. Recently reported success of DL techniques in crowd-sourced QSAR and predictive toxicology competitions has showcased these methods as powerful tools in drug-discovery and toxicology research. The aim of this work was dual, first large number of hyper-parameter configurations were explored to investigate how they affect the performance of DNNs and could act as starting points when tuning DNNs and second their performance was compared to popular methods widely employed in the field of cheminformatics namely Naïve Bayes, k-nearest neighbor, random forest and support vector machines. Moreover, robustness of machine learning methods to different levels of artificially introduced noise was assessed. The open-source Caffe deep-learning framework and modern NVidia GPU units were utilized to carry out this study, allowing large number of DNN configurations to be explored.ResultsWe show that feed-forward deep neural networks are capable of achieving strong classification performance and outperform shallow methods across diverse activity classes when optimized. Hyper-parameters that were found to play critical role are the activation function, dropout regularization, number hidden layers and number of neurons. When compared to the rest methods, tuned DNNs were found to statistically outperform, with p value <0.01 based on Wilcoxon statistical test. DNN achieved on average MCC units of 0.149 higher than NB, 0.092 than kNN, 0.052 than SVM with linear kernel, 0.021 than RF and finally 0.009 higher than SVM with radial basis function kernel. When exploring robustness to noise, non-linear methods were found to perform well when dealing with low levels of noise, lower than or equal to 20%, however when dealing with higher levels of noise, higher than 30%, the Naïve Bayes method was found to perform well and even outperform at the highest level of noise 50% more sophisticated methods across several datasets.

A Deep Convolution Neural Network Model for Vehicle Recognition and Face Recognition

In recent years, vehicle recognition has become an important application in intelligent traffic monitoring and management. In this paper, we proposed a deep convolution neural network which is no less than nine layers. A vehicle data set is employed which is collected from multiple perspectives and the deep learning framework Caffe is used to verify the proposed algorithm. Comparing with traditional vehicle recognition based on machine learning which needs vehicle location and has low accuracy of shortcomings, the proposed model uses deep convolution neural network has a better performance.

Finger vein recognition based on convolutional neural network

Biometric Authentication Technology has been widely used in this information age. As one of the most important technology of authentication, finger vein recognition attracts our attention because of its high security, reliable accuracy and excellent performance. However, the current finger vein recognition system is difficult to be applied widely because its complicated image pre-processing and not representative feature vectors. To solve this problem, a finger vein recognition method based on the convolution neural network (CNN) is proposed in the paper. The image samples are directly input into the CNN model to extract its feature vector so that we can make authentication by comparing the Euclidean distance between these vectors. Finally, the Deep Learning Framework Caffe is adopted to verify this method. The result shows that there are great improvements in both speed and accuracy rate compared to the previous research. And the model has nice robustness in illumination and rotation.

Drogue detection for autonomous aerial refueling based on convolutional neural networks

Drogue detection is a fundamental issue during the close docking phase of autonomous aerial refueling (AAR). To cope with this issue, a novel and effective method based on deep learning with convolutional neural networks (CNNs) is proposed. In order to ensure its robustness and wide application, a deep learning dataset of images was prepared by utilizing real data of “Probe and Drogue” aerial refueling, which contains diverse drogues in various environmental conditions without artificial features placed on the drogues. By employing deep learning ideas and graphics processing units (GPUs), a model for drogue detection using a Caffe deep learning framework with CNNs was designed to ensure the method’s accuracy and real-time performance. Experiments were conducted to demonstrate the effectiveness of the proposed method, and results based on real AAR data compare its performance to other methods, validating the accuracy, speed, and robustness of its drogue detection ability.