scholarly journals CNN-Cert: An Efficient Framework for Certifying Robustness of Convolutional Neural Networks

2019 ◽  
Vol 33 ◽  
pp. 3240-3247 ◽  
Author(s):  
Akhilan Boopathy ◽  
Tsui-Wei Weng ◽  
Pin-Yu Chen ◽  
Sijia Liu ◽  
Luca Daniel
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
State Of The Art ◽  
Multilayer Perceptrons ◽  
Speed Up ◽  
General Activation ◽  
Np Complete ◽  
Neural Network Classifiers ◽  
Fully Connected ◽  
Minimum Distortion

Verifying robustness of neural network classifiers has attracted great interests and attention due to the success of deep neural networks and their unexpected vulnerability to adversarial perturbations. Although finding minimum adversarial distortion of neural networks (with ReLU activations) has been shown to be an NP-complete problem, obtaining a non-trivial lower bound of minimum distortion as a provable robustness guarantee is possible. However, most previous works only focused on simple fully-connected layers (multilayer perceptrons) and were limited to ReLU activations. This motivates us to propose a general and efficient framework, CNN-Cert, that is capable of certifying robustness on general convolutional neural networks. Our framework is general – we can handle various architectures including convolutional layers, max-pooling layers, batch normalization layer, residual blocks, as well as general activation functions; our approach is efficient – by exploiting the special structure of convolutional layers, we achieve up to 17 and 11 times of speed-up compared to the state-of-the-art certification algorithms (e.g. Fast-Lin, CROWN) and 366 times of speed-up compared to the dual-LP approach while our algorithm obtains similar or even better verification bounds. In addition, CNN-Cert generalizes state-of-the-art algorithms e.g. Fast-Lin and CROWN. We demonstrate by extensive experiments that our method outperforms state-of-the-art lowerbound-based certification algorithms in terms of both bound quality and speed.

scholarly journals Effective Plug-Ins for Reducing Inference-Latency of Spiking Convolutional Neural Networks During Inference Phase

2021 ◽  
Vol 15 ◽  
Author(s):  
Xuan Chen ◽  
Xiaopeng Yuan ◽  
Gaoming Fu ◽  
Yuanyong Luo ◽  
Tao Yue ◽  
...  
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Network Architecture ◽  
State Of The Art ◽  
Low Cost ◽  
Training Phase ◽  
Working Mechanism ◽  
Simulation Results ◽  
New Perspective ◽  
Fully Connected

Convolutional Neural Networks (CNNs) are effective and mature in the field of classification, while Spiking Neural Networks (SNNs) are energy-saving for their sparsity of data flow and event-driven working mechanism. Previous work demonstrated that CNNs can be converted into equivalent Spiking Convolutional Neural Networks (SCNNs) without obvious accuracy loss, including different functional layers such as Convolutional (Conv), Fully Connected (FC), Avg-pooling, Max-pooling, and Batch-Normalization (BN) layers. To reduce inference-latency, existing researches mainly concentrated on the normalization of weights to increase the firing rate of neurons. There are also some approaches during training phase or altering the network architecture. However, little attention has been paid on the end of inference phase. From this new perspective, this paper presents 4 stopping criterions as low-cost plug-ins to reduce the inference-latency of SCNNs. The proposed methods are validated using MATLAB and PyTorch platforms with Spiking-AlexNet for CIFAR-10 dataset and Spiking-LeNet-5 for MNIST dataset. Simulation results reveal that, compared to the state-of-the-art methods, the proposed method can shorten the average inference-latency of Spiking-AlexNet from 892 to 267 time steps (almost 3.34 times faster) with the accuracy decline from 87.95 to 87.72%. With our methods, 4 types of Spiking-LeNet-5 only need 24–70 time steps per image with the accuracy decline not more than 0.1%, while models without our methods require 52–138 time steps, almost 1.92 to 3.21 times slower than us.

Redundancy-Aware Pruning of Convolutional Neural Networks

2020 ◽  
Vol 32 (12) ◽  
pp. 2532-2556
Author(s):  
Guotian Xie
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
State Of The Art ◽  
Orthogonal Transformation ◽  
Feature Space ◽  
Intermediate Space ◽  
Pruning Method ◽  
Input And Output ◽  
Speed Up ◽  
Original Feature

Pruning is an effective way to slim and speed up convolutional neural networks. Generally previous work directly pruned neural networks in the original feature space without considering the correlation of neurons. We argue that such a way of pruning still keeps some redundancy in the pruned networks. In this letter, we proposed to prune in the intermediate space in which the correlation of neurons is eliminated. To achieve this goal, the input and output of a convolutional layer are first mapped to an intermediate space by orthogonal transformation. Then neurons are evaluated and pruned in the intermediate space. Extensive experiments have shown that our redundancy-aware pruning method surpasses state-of-the-art pruning methods on both efficiency and accuracy. Notably, using our redundancy-aware pruning method, ResNet models with three times the speed-up could achieve competitive performance with fewer floating point operations per second even compared to DenseNet.

scholarly journals Using spatial-temporal ensembles of convolutional neural networks for lumen segmentation in ureteroscopy

2021 ◽  
Author(s):  
Jorge F. Lazo ◽  
Aldo Marzullo ◽  
Sara Moccia ◽  
Michele Catellani ◽  
Benoit Rosa ◽  
...  
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
State Of The Art ◽  
Automatic Segmentation ◽  
Temporal Information ◽  
Invasive Technique ◽  
Dice Similarity Coefficient ◽  
Specular Reflections ◽  
Lumen Segmentation ◽  
Previous State

Abstract Purpose Ureteroscopy is an efficient endoscopic minimally invasive technique for the diagnosis and treatment of upper tract urothelial carcinoma. During ureteroscopy, the automatic segmentation of the hollow lumen is of primary importance, since it indicates the path that the endoscope should follow. In order to obtain an accurate segmentation of the hollow lumen, this paper presents an automatic method based on convolutional neural networks (CNNs). Methods The proposed method is based on an ensemble of 4 parallel CNNs to simultaneously process single and multi-frame information. Of these, two architectures are taken as core-models, namely U-Net based in residual blocks ($$m_1$$ m 1 ) and Mask-RCNN ($$m_2$$ m 2 ), which are fed with single still-frames I(t). The other two models ($$M_1$$ M 1 , $$M_2$$ M 2 ) are modifications of the former ones consisting on the addition of a stage which makes use of 3D convolutions to process temporal information. $$M_1$$ M 1 , $$M_2$$ M 2 are fed with triplets of frames ($$I(t-1)$$ I ( t - 1 ) , I(t), $$I(t+1)$$ I ( t + 1 ) ) to produce the segmentation for I(t). Results The proposed method was evaluated using a custom dataset of 11 videos (2673 frames) which were collected and manually annotated from 6 patients. We obtain a Dice similarity coefficient of 0.80, outperforming previous state-of-the-art methods. Conclusion The obtained results show that spatial-temporal information can be effectively exploited by the ensemble model to improve hollow lumen segmentation in ureteroscopic images. The method is effective also in the presence of poor visibility, occasional bleeding, or specular reflections.

Pretrained Convolutional Neural Networks Perform Well in a Challenging Test Case: Identification of Plant Bugs (Hemiptera: Miridae) Using a Small Number of Training Images

2021 ◽  
Vol 5 (2) ◽  
Author(s):  
Alexander Knyshov ◽  
Samantha Hoang ◽  
Christiane Weirauch
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Training Image ◽  
Species Level ◽  
Identification Accuracy ◽  
Support Vector ◽  
Test Case ◽  
Learning Approaches ◽  
Lower Accuracy ◽  
Neural Network Classifiers

Abstract Automated insect identification systems have been explored for more than two decades but have only recently started to take advantage of powerful and versatile convolutional neural networks (CNNs). While typical CNN applications still require large training image datasets with hundreds of images per taxon, pretrained CNNs recently have been shown to be highly accurate, while being trained on much smaller datasets. We here evaluate the performance of CNN-based machine learning approaches in identifying three curated species-level dorsal habitus datasets for Miridae, the plant bugs. Miridae are of economic importance, but species-level identifications are challenging and typically rely on information other than dorsal habitus (e.g., host plants, locality, genitalic structures). Each dataset contained 2–6 species and 126–246 images in total, with a mean of only 32 images per species for the most difficult dataset. We find that closely related species of plant bugs can be identified with 80–90% accuracy based on their dorsal habitus alone. The pretrained CNN performed 10–20% better than a taxon expert who had access to the same dorsal habitus images. We find that feature extraction protocols (selection and combination of blocks of CNN layers) impact identification accuracy much more than the classifying mechanism (support vector machine and deep neural network classifiers). While our network has much lower accuracy on photographs of live insects (62%), overall results confirm that a pretrained CNN can be straightforwardly adapted to collection-based images for a new taxonomic group and successfully extract relevant features to classify insect species.

A novel structured sparse fully connected layer in convolutional neural networks

2021 ◽  
Author(s):  
Naoki Matsumura ◽  
Yasuaki Ito ◽  
Koji Nakano ◽  
Akihiko Kasagi ◽  
Tsuguchika Tabaru
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Fully Connected

scholarly journals Structural Damage Identification of Composite Rotors Based on Fully Connected Neural Networks and Convolutional Neural Networks

Sensors ◽  
2021 ◽  
Vol 21 (6) ◽  
pp. 2005
Author(s):  
Veronika Scholz ◽  
Peter Winkler ◽  
Andreas Hornig ◽  
Maik Gude ◽  
Angelos Filippatos
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Composite Structures ◽  
Damage Identification ◽  
Vibration Response ◽  
Damage Initiation ◽  
Matrix Cracks ◽  
Operational Life ◽  
Multiple Data Sets ◽  
Fully Connected

Damage identification of composite structures is a major ongoing challenge for a secure operational life-cycle due to the complex, gradual damage behaviour of composite materials. Especially for composite rotors in aero-engines and wind-turbines, a cost-intensive maintenance service has to be performed in order to avoid critical failure. A major advantage of composite structures is that they are able to safely operate after damage initiation and under ongoing damage propagation. Therefore, a robust, efficient diagnostic damage identification method would allow monitoring the damage process with intervention occurring only when necessary. This study investigates the structural vibration response of composite rotors by applying machine learning methods and the ability to identify, localise and quantify the present damage. To this end, multiple fully connected neural networks and convolutional neural networks were trained on vibration response spectra from damaged composite rotors with barely visible damage, mostly matrix cracks and local delaminations using dimensionality reduction and data augmentation. A databank containing 720 simulated test cases with different damage states is used as a basis for the generation of multiple data sets. The trained models are tested using k-fold cross validation and they are evaluated based on the sensitivity, specificity and accuracy. Convolutional neural networks perform slightly better providing a performance accuracy of up to 99.3% for the damage localisation and quantification.

scholarly journals Image Classification for the Automatic Feature Extraction in Human Worn Fashion Data

Mathematics ◽  
2021 ◽  
Vol 9 (6) ◽  
pp. 624
Author(s):  
Stefan Rohrmanstorfer ◽  
Mikhail Komarov ◽  
Felix Mödritscher
Keyword(s):  
Neural Networks ◽  
Feature Extraction ◽  
Image Classification ◽  
Convolutional Neural Networks ◽  
Data Augmentation ◽  
State Of The Art ◽  
Image Data ◽  
Classification Model ◽  
Upper Body ◽  
Automatic Feature Extraction

With the always increasing amount of image data, it has become a necessity to automatically look for and process information in these images. As fashion is captured in images, the fashion sector provides the perfect foundation to be supported by the integration of a service or application that is built on an image classification model. In this article, the state of the art for image classification is analyzed and discussed. Based on the elaborated knowledge, four different approaches will be implemented to successfully extract features out of fashion data. For this purpose, a human-worn fashion dataset with 2567 images was created, but it was significantly enlarged by the performed image operations. The results show that convolutional neural networks are the undisputed standard for classifying images, and that TensorFlow is the best library to build them. Moreover, through the introduction of dropout layers, data augmentation and transfer learning, model overfitting was successfully prevented, and it was possible to incrementally improve the validation accuracy of the created dataset from an initial 69% to a final validation accuracy of 84%. More distinct apparel like trousers, shoes and hats were better classified than other upper body clothes.

scholarly journals Canopy Height Estimation at Landsat Resolution Using Convolutional Neural Networks

2020 ◽  
Vol 2 (1) ◽  
pp. 23-36
Author(s):  
Syed Aamir Ali Shah ◽  
Muhammad Asif Manzoor ◽  
Abdul Bais
Keyword(s):  
Neural Networks ◽  
Random Forest ◽  
Convolutional Neural Networks ◽  
Forest Structure ◽  
State Of The Art ◽  
Spatial Association ◽  
Canopy Height ◽  
Landsat Images ◽  
Structure Estimation ◽  
Vegetation Height

Forest structure estimation is very important in geological, ecological and environmental studies. It provides the basis for the carbon stock estimation and effective means of sequestration of carbon sources and sinks. Multiple parameters are used to estimate the forest structure like above ground biomass, leaf area index and diameter at breast height. Among all these parameters, vegetation height has unique standing. In addition to forest structure estimation it provides the insight into long term historical changes and the estimates of stand age of the forests as well. There are multiple techniques available to estimate the canopy height. Light detection and ranging (LiDAR) based methods, being the accurate and useful ones, are very expensive to obtain and have no global coverage. There is a need to establish a mechanism to estimate the canopy height using freely available satellite imagery like Landsat images. Multiple studies are available which contribute in this area. The majority use Landsat images with random forest models. Although random forest based models are widely used in remote sensing applications, they lack the ability to utilize the spatial association of neighboring pixels in modeling process. In this research work, we define Convolutional Neural Network based model and analyze that model for three test configurations. We replicate the random forest based setup of Grant et al., which is a similar state-of-the-art study, and compare our results and show that the convolutional neural networks (CNN) based models not only capture the spatial association of neighboring pixels but also outperform the state-of-the-art.

scholarly journals Driving Stress Detection Using Multimodal Convolutional Neural Networks with Nonlinear Representation of Short-Term Physiological Signals

Sensors ◽  
2021 ◽  
Vol 21 (7) ◽  
pp. 2381
Author(s):  
Jaewon Lee ◽  
Hyeonjeong Lee ◽  
Miyoung Shin
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Traffic Accidents ◽  
Physiological Signals ◽  
Superior Performance ◽  
Short Term ◽  
Input Signals ◽  
Skin Response ◽  
Novel Method ◽  
Fully Connected

Mental stress can lead to traffic accidents by reducing a driver’s concentration or increasing fatigue while driving. In recent years, demand for methods to detect drivers’ stress in advance to prevent dangerous situations increased. Thus, we propose a novel method for detecting driving stress using nonlinear representations of short-term (30 s or less) physiological signals for multimodal convolutional neural networks (CNNs). Specifically, from hand/foot galvanic skin response (HGSR, FGSR) and heart rate (HR) short-term input signals, first, we generate corresponding two-dimensional nonlinear representations called continuous recurrence plots (Cont-RPs). Second, from the Cont-RPs, we use multimodal CNNs to automatically extract FGSR, HGSR, and HR signal representative features that can effectively differentiate between stressed and relaxed states. Lastly, we concatenate the three extracted features into one integrated representation vector, which we feed to a fully connected layer to perform classification. For the evaluation, we use a public stress dataset collected from actual driving environments. Experimental results show that the proposed method demonstrates superior performance for 30-s signals, with an overall accuracy of 95.67%, an approximately 2.5–3% improvement compared with that of previous works. Additionally, for 10-s signals, the proposed method achieves 92.33% classification accuracy, which is similar to or better than the performance of other methods using long-term signals (over 100 s).

scholarly journals Segment convolutional neural networks (Seg-CNNs) for classifying relations in clinical notes

2017 ◽  
Vol 25 (1) ◽  
pp. 93-98 ◽  
Author(s):  
Yuan Luo ◽  
Yu Cheng ◽  
Özlem Uzuner ◽  
Peter Szolovits ◽  
Justin Starren
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
State Of The Art ◽  
Graphics Processing Unit ◽  
Medical Problem ◽  
Feature Engineering ◽  
Processing Unit ◽  
Clinical Notes ◽  
Overall Evaluation ◽  
Relation Classification

Abstract We propose Segment Convolutional Neural Networks (Seg-CNNs) for classifying relations from clinical notes. Seg-CNNs use only word-embedding features without manual feature engineering. Unlike typical CNN models, relations between 2 concepts are identified by simultaneously learning separate representations for text segments in a sentence: preceding, concept1, middle, concept2, and succeeding. We evaluate Seg-CNN on the i2b2/VA relation classification challenge dataset. We show that Seg-CNN achieves a state-of-the-art micro-average F-measure of 0.742 for overall evaluation, 0.686 for classifying medical problem–treatment relations, 0.820 for medical problem–test relations, and 0.702 for medical problem–medical problem relations. We demonstrate the benefits of learning segment-level representations. We show that medical domain word embeddings help improve relation classification. Seg-CNNs can be trained quickly for the i2b2/VA dataset on a graphics processing unit (GPU) platform. These results support the use of CNNs computed over segments of text for classifying medical relations, as they show state-of-the-art performance while requiring no manual feature engineering.

Sign in / Sign up

Export Citation Format

Share Document