基于CNN-GAP可解释性模型的软件源码漏洞检测方法

源代码漏洞检测是保证软件系统安全的重要手段。近年来,多种深度学习模型应用于源代码漏洞检测,极大提高了漏洞检测的效率,但还存在自定义标识符导致库外词过多、嵌入词向量的语义不够准确、神经网络模型缺乏可解释性等问题。基于此,该文提出了一种基于卷积神经网络(CNN)和全局平均池化(GAP)可解释性模型的源代码漏洞检测方法。首先在源代码预处理中对部分自定义标识符进行归一化,并采用One-hot编码进行词嵌入以缓解库外词过多的问题;然后构建CNN-GAP神经网络模型,识别出包含CWE-119缓冲区溢出类型漏洞的函数;最后通过类激活映射(CAM)可解释方法对结果进行可视化输出,标识出可能与漏洞相关的代码。通过与Russell等人提出的模型以及Li等人提出的VulDeePecker模型进行对比分析,表明CNN-GAP模型能达到相当甚至更好的性能,且具有一定的可解释性,便于研究人员对漏洞进行更深入的分析。     

A novel deep learning framework for double JPEG compression detection of small size blocks

Double JPEG compression detection plays a vital role in multimedia forensics, to find out whether a JPEG image is authentic or manipulated. However, it still remains to be a challenging task in the case when the quality factor of the first compression is much higher than that of the second compression, as well as in the case when the targeted image blocks are quite small. In this work, we present a novel end-to-end deep learning framework taking raw DCT coefficients as input to distinguish between single and double compressed images, which performs superior in the above two cases. Our proposed framework can be divided into two stages. In the first stage, we adopt an auxiliary DCT layer with sixty-four 8 × 8 DCT kernels. Using a specific layer to extract DCT coefficients instead of extracting them directly from JPEG bitstream allows our proposed framework to work even if the double compressed images are stored in spatial domain, e.g. in PGM, TIFF or other bitmap formats. The second stage is a deep neural network with multiple convolutional blocks to extract more effective features. We have conducted extensive experiments on three different image datasets. The experimental results demonstrate the superiority of our framework when compared with other state-of-the-art double JPEG compression detection methods either hand-crafted or learned using deep networks in the literature, especially in the two cases mentioned above. Furthermore, our proposed framework can detect triple and even multiple JPEG compressed images, which is scarce in the literature as far as we know.     

基于增量式深度神经网络的图像分类系统

为解决图像分类任务中模型结构固化、产生巨大内存消耗、时间消耗的问题,提出一种增量式深度神经网络（IDNN）。输入样本通过聚类算法激活不同簇并被分别处理：如果新样本激活已有簇,则更新该簇参数;否则为新簇开辟分支,并训练独立特征集。在Caltech-101、ORL Face、ETH-80数据库的验证结果表明,该系统能自动调整网络结构,适用于轮廓、纹理、视角等不同环境的增量式学习,例如在Caltech-101库分类任务中准确率超出VGGNet 5.08%、AlexNet 3.44%。     

Towards improving detection performance for malware with a correntropy-based deep learning method

With the rapid development of Internet of Things (IoT) technologies, the detection and analysis of malware have become a matter of concern in the industrial application of Cyber-Physical System (CPS) that provides various services using the IoT paradigm. Currently, many advanced machine learning methods such as deep learning are popular in the research of malware detection and analysis, and some achievements have been made so far. However, there are also some problems. For example, considering the noise and outliers in the existing datasets of malware, some methods are not robust enough. Therefore, the accuracy of malware classification still needs to be improved. Aiming at this issue, we propose a novel method that combines the correntropy and the deep learning model. In our proposed method for malware detection and analysis, given the success of the mixture correntropy as an effective similarity measure in addressing complex datasets with noise, it is therefore incorporated into a popular deep learning model, i.e., Convolutional Neural Network (CNN), to reconstruct its loss function, with the purpose of further detecting the features of outliers. We present the detailed design process of our method. Furthermore, the proposed method is tested both on a real-world malware dataset and a popular benchmark dataset to verify its learning performance.     

EEG-based prediction of driver's cognitive performance by deep convolutional neural network

We considered the prediction of driver's cognitive states related to driving performance using EEG signals. We proposed a novel channel-wise convolutional neural network (CCNN) whose architecture considers the unique characteristics of EEG data. We also discussed CCNN-R, a CCNN variation that uses Restricted Boltzmann Machine to replace the convolutional filter, and derived the detailed algorithm. To test the performance of CCNN and CCNN-R, we assembled a large EEG dataset from 3 studies of driver fatigue that includes samples from 37 subjects. Using this dataset, we investigated the new CCNN and CCNN-R on raw EEG data and also Independent Component Analysis (ICA) decomposition. We tested both within-subject and cross-subject predictions and the results showed CCNN and CCNN-R achieved robust and improved performance over conventional DNN and CNN as well as other non-DL algorithms.     

Robust object representation by boosting-like deep learning architecture

This paper presents a new deep learning architecture for robust object representation, aiming at efficiently combining the proposed synchronized multi-stage feature (SMF) and a boosting-like algorithm. The SMF structure can capture a variety of characteristics from the inputting object based on the fusion of the handcraft features and deep learned features. With the proposed boosting-like algorithm, we can obtain more convergence stability on training multi-layer network by using the boosted samples. We show the generalization of our object representation architecture by applying it to undertake various tasks, i.e. pedestrian detection and action recognition. Our approach achieves 15.89% and 3.85% reduction in the average miss rate compared with ACF and JointDeep on the largest Caltech dataset, and acquires competitive results on the MSRAction3D dataset.     

A new hardware Trojan detection technique using deep convolutional neural network

The involvement of external vendors in semiconductor industries increases the chance of hardware Trojan (HT) insertion in different phases of the integrated circuit (IC) design. Recently, several partial reverse engineering (RE) based HT detection techniques are reported, which attempt to reduce the time and complexity involved in the full RE process by applying machine learning or image processing techniques in IC images. However, these techniques fail to extract the relevant image features, not robust to image variations, complicated, less generalizable, and possess a low detection rate. Therefore, to overcome the above limitations, this paper proposes a new partial RE based HT detection technique that detects Trojans from IC layout images using Deep Convolutional Neural Network (DCNN). The proposed DCNN model consists of stacking several convolutional and pooling layers. It layer-wise extracts and selects the most relevant and robust features automatically from the IC images and eliminates the need to apply the feature extraction algorithm separately. To prevent the over-training of the DCNN model, a new stopping condition method and two new metrics, namely Accuracy difference measure (ADM) and Loss difference measure (LDM), are proposed that halts the training only when the performance of our model genuinely drops. Further, to combat the issue of process variations and fabrication noise generated during the RE process, we include noisy images with varying parameters in the training process of the model. We also apply the data augmentation and regularization techniques in the model to address the issues of underfitting and overfitting. Experimental evaluation shows that the proposed technique provides 99% and 97.4% accuracy on Trust-Hub and synthetic ISCAS dataset, respectively, which is on-an-average 15.83% and 21.69% higher than the existing partial RE based techniques.     

Normalization and dropout for stochastic computing-based deep convolutional neural networks

Recently, Deep Convolutional Neural Network (DCNN) has been recognized as the most effective model for pattern recognition and classification tasks. With the fast growing Internet of Things (IoTs) and wearable devices, it becomes attractive to implement DCNNs in embedded and portable systems. However, novel computing paradigms are urgently required to deploy DCNNs that have huge power consumptions and complex topologies in systems with limited area and power supply. Recent works have demonstrated that Stochastic Computing (SC) can radically simplify the hardware implementation of arithmetic units and has the potential to bring the success of DCNNs to embedded systems. This paper introduces normalization and dropout, which are essential techniques for the state-of-the-art DCNNs, to the existing SC-based DCNN frameworks. In this work, the feature extraction block of DCNNs is implemented using an approximate parallel counter, a near-max pooling block and an SC-based rectified linear activation unit. A novel SC-based normalization design is proposed, which includes a square and summation unit, an activation unit and a division unit. The dropout technique is integrated into the training phase and the learned weights are adjusted during the hardware implementation. Experimental results on AlexNet with the ImageNet dataset show that the SC-based DCNN with the proposed normalization and dropout techniques achieves 3.26% top-1 accuracy improvement and 3.05% top-5 accuracy improvement compared with the SC-based DCNN without these two essential techniques, confirming the effectiveness of our normalization and dropout designs.     

Video frame interpolation using deep cascaded network structure

Video frame interpolation is a technology that generates high frame rate videos from low frame rate videos by using the correlation between consecutive frames. Presently, convolutional neural networks (CNN) exhibit outstanding performance in image processing and computer vision. Many variant methods of CNN have been proposed for video frame interpolation by estimating either dense motion flows or kernels for moving objects. However, most methods focus on estimating accurate motion. In this study, we exhaustively analyze the advantages of both motion estimation schemes and propose a cascaded system to maximize the advantages of both the schemes. The proposed cascaded network consists of three autoencoder networks, that process the initial frame interpolation and its refinement. The quantitative and qualitative evaluations demonstrate that the proposed cascaded structure exhibits a promising performance compared to currently existing state-of-the-art-methods.     

Image super-resolution reconstruction based on sparse representation and deep learning

Super-resolution reconstruction technology has important scientific significance and application value in the field of image processing by performing image restoration processing on one or more low-resolution images to improve image spatial resolution. Based on the SCSR algorithm and VDSR network, in order to further improve the image reconstruction quality, an image super-resolution reconstruction algorithm combined with multi-residual network and multi-feature SCSR(MRMFSCSR) is proposed. Firstly, at the sparse reconstruction stage, according to the characteristics of image blocks, our algorithm extracts the contour features of non-flat blocks by NSCT transform, extracts the texture features of flat blocks by Gabor transform, then obtains the reconstructed high-resolution (HR) images by using sparse models. Secondly, according to improve the VDSR deep network and introduce the feature fusion idea, the multi-residual network structure (MR) is designed. The reconstructed HR image obtained by the sparse reconstruction stage is used as the input of the MR network structure to optimize the high-frequency detail residual information. Finally, we can obtain a higher quality super-resolution image compared with the SCSR algorithm and the VDSR algorithm.     

INDUCTION OF DECISION TREES BASED ON A FUZZY NEURAL NETWORK

Based on a fuzzy neural network, the letter presents an approach for the induction of decision trees.The approach makes use of the weights of fuzzy mappings in the fuzzy neural network which has been trained.It can realize the optimization of fuzzy decision trees by branch cutting, and improve the ratio of correctness and efficiency of the induction of decision trees.     

Hand pose estimation in object-interaction based on deep learning for virtual reality applications

Hand Pose Estimation aims to predict the position of joints on a hand from an image, and it has become popular because of the emergence of VR/AR/MR technology. Nevertheless, an issue surfaces when trying to achieve this goal, since a hand tends to cause self-occlusion or external occlusion easily as it interacts with external objects. As a result, there have been many projects dedicated to this field for a better solution of this problem. This paper develops a system that accurately estimates a hand pose in 3D space using depth images for VR applications. We propose a data-driven approach of training a deep learning model for hand pose estimation with object interaction. In the convolutional neural network (CNN) training procedure, we design a skeleton-difference loss function, which effectively can learn the physical constraints of a hand. Also, we propose an object-manipulating loss function, which considers knowledge of the hand-object interaction, to enhance performance.In the experiments we have conducted for hand pose estimation under different conditions, the results validate the robustness and the performance of our system and show that our method is able to predict the joints more accurately in challenging environmental settings. Such appealing results may be attributed to the consideration of the physical joint relationship as well as object information, which in turn can be applied to future VR/AR/MR systems for more natural experience.     

Context prior-based with residual learning for face detection: A deep convolutional encoder–decoder network

In the field of security, faces are usually blurry, occluded, diverse pose and small in the image captured by an outdoor surveillance camera, which is affected by the external environment such as the camera pose and range, weather conditions, etc. It can be described as a problem of hard face detection in natural images. To solve this problem, we propose a deep convolutional neural network named feature hierarchy encoder–decoder network (FHEDN). It is motivated by two observations from contextual semantic information and the mechanism of multi-scale face detection. The proposed network is a scale-variant style architecture and single stage, which are composed of encoder and decoder subnetworks. Based on the assumption that contextual semantic information around face being auxiliary to detect faces, we introduce a residual mechanism to fuse context prior-based information into face feature and formulate the learning chain to train each encoder–decoder pair. In addition, we discuss some important factors in implement details such as the distribution of training dataset, the scale of feature hierarchy, and anchor box size, etc. They have some impact on the detection performance of the final network. Compared with some state-of-the-art algorithms, our method achieves promising performance on the popular benchmarks including AFW, PASCAL FACE, FDDB, and WIDER FACE. Consequently, the proposed approach can be efficiently implemented and routinely applied to detect faces with severe occlusion and arbitrary pose variations in unconstrained scenes. Our code and results are available on https://github.com/zzxcoder/EvaluationFHEDN.     

Structure-prior deep neural network for lane detection

Lane detection is an important task of road environment perception for autonomous driving. Deep learning methods based on semantic segmentation have been successfully applied to lane detection, but they require considerable computational cost for high complexity. The lane detection is treated as a particular semantic segmentation task due to the prior structural information of lane markings which have long continuous shape. Most traditional CNN are designed for the representation learning of semantic information, while this prior structural information is not fully exploited. In this paper, we propose a recurrent slice convolution module (called RSCM) to exploit the prior structural information of lane markings. The proposed RSCM is a special recurrent network structure with several slice convolution units (called SCU). The RSCM could obtain stronger semantic representation through the propagation of the prior structural information in SCU. Furthermore, we design a distance loss in consideration of the prior structure of lane markings. The lane detection network can be trained more steadily via the overall loss function formed by combining segmentation loss with the distance loss. The experimental results show the effectiveness of our method. We achieve excellent computation efficiency while keeping decent detection quality on lane detection benchmarks and the computational cost of our method is much lower than the state-of-the-art methods.     

基于模糊神经网的决策树生成

数据挖掘技术能从大量数据中挖掘和发现有价值和隐含的知识,因而得到广泛的研究和应用。提出了一种基于五层模糊神经网络的决策树生成方法：首先运用五层模糊神经网络学习变量间的模糊映射关系。然后从中生成模糊决策树。这种方法利用了五层模糊神经网络学习的模糊映射强度,并能实现模糊决策树的剪枝优化,提高了算法的正确率和效率。     

Real-time license plate detection and recognition using deep convolutional neural networks

Automatic License Plate Recognition (ALPR) is an important task with many applications in Intelligent Transportation and Surveillance systems. This work presents an end-to-end ALPR method based on a hierarchical Convolutional Neural Network (CNN). The core idea of the proposed method is to identify the vehicle and the license plate region using two passes on the same CNN, and then to recognize the characters using a second CNN. The recognition CNN massively explores the use of synthetic and augmented data to cope with limited training datasets, and our results show that the augmentation process significantly increases the recognition rate. In addition, we present a novel temporal coherence technique to better stabilize the OCR output in videos. Our method was tested with publicly available datasets containing Brazilian and European license plates, achieving accuracy rates better than competitive academic methods and a commercial system.     

Nrityabodha: Towards understanding Indian classical dance using a deep learning approach

Indian classical dance has existed since over 5000 years and is widely practised and performed all over the world. However, the semantic meaning of the dance gestures and body postures as well as the intricate steps accompanied by music and recital of poems is only understood fully by the connoisseur. The common masses who watch a concert rarely appreciate or understand the ideas conveyed by the dancer. Can machine learning algorithms aid a novice to understand the semantic intricacies being expertly conveyed by the dancer? In this work, we aim to address this highly challenging problem and propose deep learning based algorithms to identify body postures and hand gestures in order to comprehend the intended meaning of the dance performance. Specifically, we propose a convolutional neural network and validate its performance on standard datasets for poses and hand gestures as well as on constrained and real-world datasets of classical dance. We use transfer learning to show that the pre-trained deep networks can reduce the time taken during training and also improve accuracy. Interestingly, we show with experiments performed using Kinect in constrained laboratory settings and data from Youtube that it is possible to identify body poses and hand gestures of the performer to understand the semantic meaning of the enacted dance piece.