scholarly journals Detecting Local Insights from Global Labels: Supervised & Zero-Shot Sequence Labeling via a Convolutional Decomposition

2021 ◽  
pp. 1-46
Author(s):  
Allen Schmaltz
Keyword(s):  
Neural Network ◽  
Feature Detection ◽  
Decision Rules ◽  
Binary Sequence ◽  
Ground Truth ◽  
K Nearest Neighbors ◽  
Deep Network ◽  
Sequence Labeling ◽  
Support Set ◽  
Linear Layer

Abstract We propose a new, more actionable view of neural network interpretability and data analysis by leveraging the remarkable matching effectiveness of representations derived from deep networks, guided by an approach for class-conditional feature detection. The decomposition of the filterngram interactions of a convolutional neural network and a linear layer over a pre-trained deep network yields a strong binary sequence labeler, with flexibility in producing predictions at— and defining loss functions for—varying label granularities, from the fully-supervised sequence labeling setting to the challenging zero-shot sequence labeling setting, in which we seek tokenlevel predictions but only have document-level labels for training. From this sequence-labeling layer we derive dense representations of the input that can then be matched to instances from training, or a support set with known labels. Such introspection with inference-time decision rules provides a means, in some settings, of making local updates to the model by altering the labels or instances in the support set without re-training the full model. Finally, we construct a particular K-nearest neighbors (K-NN) model from matched exemplar representations that approximates the original model’s predictions and is at least as effective a predictor with respect to the ground-truth labels. This additionally yields interpretable heuristics at the token level for determining when predictions are less likely to be reliable, and for screening input dissimilar to the support set. In effect, we show that we can transform the deep network into a simple weighting over exemplars and associated labels, yielding an introspectable—and modestly updatable— version of the original model.

Carpal Bone Segmentation Using Fully Convolutional Neural Network

2019 ◽  
Vol 15 (10) ◽  
pp. 983-989
Author(s):  
Liang Kim Meng ◽  
Azira Khalil ◽  
Muhamad Hanif Ahmad Nizar ◽  
Maryam Kamarun Nisham ◽  
Belinda Pingguan-Murphy ◽  
...  
Keyword(s):  
Neural Network ◽  
Convolutional Neural Network ◽  
Automatic Segmentation ◽  
Ground Truth ◽  
Bone Age ◽  
Image Feature ◽  
Qualitative Assessment ◽  
Carpal Bone ◽  
Radiographic Images ◽  
Accurate Quantitative Analysis

Background: Bone Age Assessment (BAA) refers to a clinical procedure that aims to identify a discrepancy between biological and chronological age of an individual by assessing the bone age growth. Currently, there are two main methods of executing BAA which are known as Greulich-Pyle and Tanner-Whitehouse techniques. Both techniques involve a manual and qualitative assessment of hand and wrist radiographs, resulting in intra and inter-operator variability accuracy and time-consuming. An automatic segmentation can be applied to the radiographs, providing the physician with more accurate delineation of the carpal bone and accurate quantitative analysis. Methods: In this study, we proposed an image feature extraction technique based on image segmentation with the fully convolutional neural network with eight stride pixel (FCN-8). A total of 290 radiographic images including both female and the male subject of age ranging from 0 to 18 were manually segmented and trained using FCN-8. Results and Conclusion: The results exhibit a high training accuracy value of 99.68% and a loss rate of 0.008619 for 50 epochs of training. The experiments compared 58 images against the gold standard ground truth images. The accuracy of our fully automated segmentation technique is 0.78 ± 0.06, 1.56 ±0.30 mm and 98.02% in terms of Dice Coefficient, Hausdorff Distance, and overall qualitative carpal recognition accuracy, respectively.

scholarly journals Geometric property-based convolutional neural network for indoor object detection

2021 ◽  
Vol 18 (1) ◽  
pp. 172988142199332
Author(s):  
Xintao Ding ◽  
Boquan Li ◽  
Jinbao Wang
Keyword(s):  
Neural Network ◽  
Object Detection ◽  
Convolutional Neural Network ◽  
Geometric Property ◽  
Ground Truth ◽  
Geometric Constraints ◽  
Depth Information ◽  
Training Set ◽  
Object Knowledge ◽  
The Mean

Indoor object detection is a very demanding and important task for robot applications. Object knowledge, such as two-dimensional (2D) shape and depth information, may be helpful for detection. In this article, we focus on region-based convolutional neural network (CNN) detector and propose a geometric property-based Faster R-CNN method (GP-Faster) for indoor object detection. GP-Faster incorporates geometric property in Faster R-CNN to improve the detection performance. In detail, we first use mesh grids that are the intersections of direct and inverse proportion functions to generate appropriate anchors for indoor objects. After the anchors are regressed to the regions of interest produced by a region proposal network (RPN-RoIs), we then use 2D geometric constraints to refine the RPN-RoIs, in which the 2D constraint of every classification is a convex hull region enclosing the width and height coordinates of the ground-truth boxes on the training set. Comparison experiments are implemented on two indoor datasets SUN2012 and NYUv2. Since the depth information is available in NYUv2, we involve depth constraints in GP-Faster and propose 3D geometric property-based Faster R-CNN (DGP-Faster) on NYUv2. The experimental results show that both GP-Faster and DGP-Faster increase the performance of the mean average precision.

scholarly journals Classification of Cattle Behaviours Using Neck-Mounted Accelerometer-Equipped Collars and Convolutional Neural Networks

Sensors ◽  
2021 ◽  
Vol 21 (12) ◽  
pp. 4050
Author(s):  
Dejan Pavlovic ◽  
Christopher Davison ◽  
Andrew Hamilton ◽  
Oskar Marko ◽  
Robert Atkinson ◽  
...  
Keyword(s):  
Neural Network ◽  
Model Performance ◽  
Ground Truth ◽  
Practical Implementation ◽  
Ground Truth Data ◽  
Battery Lifetime ◽  
Implementation Challenges ◽  
Memory Footprint ◽  
Commercial Farms ◽  
Using Data

Monitoring cattle behaviour is core to the early detection of health and welfare issues and to optimise the fertility of large herds. Accelerometer-based sensor systems that provide activity profiles are now used extensively on commercial farms and have evolved to identify behaviours such as the time spent ruminating and eating at an individual animal level. Acquiring this information at scale is central to informing on-farm management decisions. The paper presents the development of a Convolutional Neural Network (CNN) that classifies cattle behavioural states (`rumination’, `eating’ and `other’) using data generated from neck-mounted accelerometer collars. During three farm trials in the United Kingdom (Easter Howgate Farm, Edinburgh, UK), 18 steers were monitored to provide raw acceleration measurements, with ground truth data provided by muzzle-mounted pressure sensor halters. A range of neural network architectures are explored and rigorous hyper-parameter searches are performed to optimise the network. The computational complexity and memory footprint of CNN models are not readily compatible with deployment on low-power processors which are both memory and energy constrained. Thus, progressive reductions of the CNN were executed with minimal loss of performance in order to address the practical implementation challenges, defining the trade-off between model performance versus computation complexity and memory footprint to permit deployment on micro-controller architectures. The proposed methodology achieves a compression of 14.30 compared to the unpruned architecture but is nevertheless able to accurately classify cattle behaviours with an overall F1 score of 0.82 for both FP32 and FP16 precision while achieving a reasonable battery lifetime in excess of 5.7 years.

scholarly journals Unraveling the deep learning gearbox in optical coherence tomography image segmentation towards explainable artificial intelligence

2021 ◽  
Vol 4 (1) ◽  
Author(s):  
Peter M. Maloca ◽  
Philipp L. Müller ◽  
Aaron Y. Lee ◽  
Adnan Tufail ◽  
Konstantinos Balaskas ◽  
...  
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Optical Coherence Tomography ◽  
Image Segmentation ◽  
Convolutional Neural Network ◽  
Learning Algorithm ◽  
Ground Truth ◽  
Optical Coherence Tomography Image ◽  
Optical Coherence ◽  
Tomography Image

AbstractMachine learning has greatly facilitated the analysis of medical data, while the internal operations usually remain intransparent. To better comprehend these opaque procedures, a convolutional neural network for optical coherence tomography image segmentation was enhanced with a Traceable Relevance Explainability (T-REX) technique. The proposed application was based on three components: ground truth generation by multiple graders, calculation of Hamming distances among graders and the machine learning algorithm, as well as a smart data visualization (‘neural recording’). An overall average variability of 1.75% between the human graders and the algorithm was found, slightly minor to 2.02% among human graders. The ambiguity in ground truth had noteworthy impact on machine learning results, which could be visualized. The convolutional neural network balanced between graders and allowed for modifiable predictions dependent on the compartment. Using the proposed T-REX setup, machine learning processes could be rendered more transparent and understandable, possibly leading to optimized applications.

A machine learning approach to estimate the strain energy absorption in expanded polystyrene foams

2021 ◽  
pp. 0021955X2110210
Author(s):  
Alejandro E Rodríguez-Sánchez ◽  
Héctor Plascencia-Mora
Keyword(s):  
Neural Network ◽  
Energy Absorption ◽  
Mechanical Energy ◽  
Compressive Loading ◽  
Ground Truth ◽  
Expanded Polystyrene ◽  
Polystyrene Foam ◽  
Stress Strain ◽  
Ground Truth Data ◽  
Expanded Polystyrene Foam

Traditional modeling of mechanical energy absorption due to compressive loadings in expanded polystyrene foams involves mathematical descriptions that are derived from stress/strain continuum mechanics models. Nevertheless, most of those models are either constrained using the strain as the only variable to work at large deformation regimes and usually neglect important parameters for energy absorption properties such as the material density or the rate of the applying load. This work presents a neural-network-based approach that produces models that are capable to map the compressive stress response and energy absorption parameters of an expanded polystyrene foam by considering its deformation, compressive loading rates, and different densities. The models are trained with ground-truth data obtained in compressive tests. Two methods to select neural network architectures are also presented, one of which is based on a Design of Experiments strategy. The results show that it is possible to obtain a single artificial neural networks model that can abstract stress and energy absorption solution spaces for the conditions studied in the material. Additionally, such a model is compared with a phenomenological model, and the results show than the neural network model outperforms it in terms of prediction capabilities, since errors around 2% of experimental data were obtained. In this sense, it is demonstrated that by following the presented approach is possible to obtain a model capable to reproduce compressive polystyrene foam stress/strain data, and consequently, to simulate its energy absorption parameters.

scholarly journals Hyperspectral Image Classification Based on Multi-Scale Residual Network with Attention Mechanism

2021 ◽  
Vol 13 (3) ◽  
pp. 335
Author(s):  
Yuhao Qing ◽  
Wenyi Liu
Keyword(s):  
Neural Network ◽  
Deep Learning ◽  
Convolutional Neural Network ◽  
Image Classification ◽  
Classification Accuracy ◽  
Hyperspectral Image ◽  
Principal Component ◽  
Hyperspectral Image Classification ◽  
Deep Network ◽  
Multi Scale

In recent years, image classification on hyperspectral imagery utilizing deep learning algorithms has attained good results. Thus, spurred by that finding and to further improve the deep learning classification accuracy, we propose a multi-scale residual convolutional neural network model fused with an efficient channel attention network (MRA-NET) that is appropriate for hyperspectral image classification. The suggested technique comprises a multi-staged architecture, where initially the spectral information of the hyperspectral image is reduced into a two-dimensional tensor, utilizing a principal component analysis (PCA) scheme. Then, the constructed low-dimensional image is input to our proposed ECA-NET deep network, which exploits the advantages of its core components, i.e., multi-scale residual structure and attention mechanisms. We evaluate the performance of the proposed MRA-NET on three public available hyperspectral datasets and demonstrate that, overall, the classification accuracy of our method is 99.82 %, 99.81%, and 99.37, respectively, which is higher compared to the corresponding accuracy of current networks such as 3D convolutional neural network (CNN), three-dimensional residual convolution structure (RES-3D-CNN), and space–spectrum joint deep network (SSRN).

scholarly journals Improving Machine Learning Identification of Unsafe Driver Behavior by Means of Sensor Fusion

2020 ◽  
Vol 10 (18) ◽  
pp. 6417 ◽  
Author(s):  
Emanuele Lattanzi ◽  
Giacomo Castellucci ◽  
Valerio Freschi
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Driver Behavior ◽  
Ground Truth ◽  
Support Vector ◽  
Svm Classifier ◽  
Learning Technology ◽  
Average Accuracy ◽  
Unsafe Behaviors ◽  
Vehicle Sensors

Most road accidents occur due to human fatigue, inattention, or drowsiness. Recently, machine learning technology has been successfully applied to identifying driving styles and recognizing unsafe behaviors starting from in-vehicle sensors signals such as vehicle and engine speed, throttle position, and engine load. In this work, we investigated the fusion of different external sensors, such as a gyroscope and a magnetometer, with in-vehicle sensors, to increase machine learning identification of unsafe driver behavior. Starting from those signals, we computed a set of features capable to accurately describe the behavior of the driver. A support vector machine and an artificial neural network were then trained and tested using several features calculated over more than 200 km of travel. The ground truth used to evaluate classification performances was obtained by means of an objective methodology based on the relationship between speed, and lateral and longitudinal acceleration of the vehicle. The classification results showed an average accuracy of about 88% using the SVM classifier and of about 90% using the neural network demonstrating the potential capability of the proposed methodology to identify unsafe driver behaviors.

scholarly journals Deep Multi-State Dynamic Recurrent Neural Networks Operating on Wavelet Based Neural Features for Robust Brain Machine Interfaces

2019 ◽  
Author(s):  
Benyamin Haghi ◽  
Spencer Kellis ◽  
Sahil Shah ◽  
Maitreyi Ashok ◽  
Luke Bashford ◽  
...  
Keyword(s):  
Neural Network ◽  
Posterior Parietal Cortex ◽  
Ground Truth ◽  
Open Loop ◽  
Neural Data ◽  
State Dynamic ◽  
History Of ◽  
Posterior Parietal ◽  
Machine Interface ◽  
Memory Accesses

AbstractWe present a new deep multi-state Dynamic Recurrent Neural Network (DRNN) architecture for Brain Machine Interface (BMI) applications. Our DRNN is used to predict Cartesian representation of a computer cursor movement kinematics from open-loop neural data recorded from the posterior parietal cortex (PPC) of a human subject in a BMI system. We design the algorithm to achieve a reasonable trade-off between performance and robustness, and we constrain memory usage in favor of future hardware implementation. We feed the predictions of the network back to the input to improve prediction performance and robustness. We apply a scheduled sampling approach to the model in order to solve a statistical distribution mismatch between the ground truth and predictions. Additionally, we configure a small DRNN to operate with a short history of input, reducing the required buffering of input data and number of memory accesses. This configuration lowers the expected power consumption in a neural network accelerator. Operating on wavelet-based neural features, we show that the average performance of DRNN surpasses other state-of-the-art methods in the literature on both single- and multi-day data recorded over 43 days. Results show that multi-state DRNN has the potential to model the nonlinear relationships between the neural data and kinematics for robust BMIs.

scholarly journals Methodology for detection of paroxysmal atrial fibrillation based on P-Wave, HRV and QR electrical alternans features

2020 ◽  
Vol 10 (4) ◽  
pp. 4023
Author(s):  
Henry Castro ◽  
Juan David Garcia-Racines ◽  
Alvaro Bernal-Noreña
Keyword(s):  
Neural Network ◽  
Atrial Fibrillation ◽  
Paroxysmal Atrial Fibrillation ◽  
P Wave ◽  
Computational Techniques ◽  
K Nearest Neighbors ◽  
Ecg Signals ◽  
Complex Process ◽  
Artificial Neural Network Ann ◽  
Electrical Alternans

The detection of Paroxysmal Atrial Fibrillation (PAF) is a fairly complex process performed manually by cardiologists or electrophysiologists by reading an electrocardiogram (ECG). Currently, computational techniques for automatic detection based on fast Fourier transform (FFT), Bayes optimal classifier (BOC), k-nearest neighbors (K-NNs), and artificial neural network (ANN) have been proposed. In this study, six features were obtained based on the morphology of the P-Wave, the QRS complex and the heart rate variability (HRV) of the ECG. The performance of this methodology was validated using clinical ECG signals from the Physionet arrhythmia database MIT-BIH. A feedforward neural network was used to detect the presence of PAF reaching a general accuracy of 97.4%. The results obtained show that the inclusion of the information of the P-Wave, HRV and QR Electrical alternans increases the accuracy to identify the PAF event compared to other works that use the information of only one or at most two of them.

Sign in / Sign up

Export Citation Format

Share Document