Imperceptible Adversarial Attacks against Traffic Scene Recognition

Mapping Intimacies ◽

10.21203/rs.3.rs-652216/v1 ◽

2021 ◽

Author(s):

Yinghui Zhu ◽

Yuzhen Jiang

Keyword(s):

Semantic Segmentation ◽

Absolute Error ◽

Scene Recognition ◽

Lane Detection ◽

Traffic Sign ◽

Adversarial Examples ◽

The Mean ◽

Street Scene ◽

Human Eyes ◽

To Receive

Abstract Adversarial examples have begun to receive widespread attention owning to their potential destructions to the most popular DNNs. They are crafted from original images by embedding well calculated perturbations. In some cases the perturbations are so slight that neither human eyes nor monitoring systems can notice easily and such imperceptibility makes them have greater concealment and damage. For the sake of investigating the invisible dangers in the applications of traffic DNNs, we focus on imperceptible adversarial attacks on different traffic vision tasks, including traffic sign classification, lane detection and street scene recognition. We propose an universal logits map-based attack architecture against image semantic segmentation, and design two targeted attack approaches on it. All the attack algorithms generate the micro-noise adversarial examples by the iterative method of gradient descent optimization. All of them can achieve 100% attack rate but with very low distortion, among which, the mean MAE (Mean Absolute Error) of perturbation noise based on traffic sign classifier attack is as low as 0.562, and the other two algorithms based on semantic segmentation are only 1.574 and 1.503. We believe that our research on imperceptible adversarial attacks can be of substantial reference to the security of DNNs applications.

Download Full-text

The Antiaggressive Action of Combined Haloperidol-Propranolol Treatment in Schizophrenia

European Psychologist ◽

10.1027//1016-9040.5.4.312 ◽

2000 ◽

Vol 5 (4) ◽

pp. 312-325 ◽

Cited By ~ 9

Author(s):

Gadi Maoz ◽

Daniel Stein ◽

Sorin Meged ◽

Larisa Kurzman ◽

Joseph Levine ◽

...

Keyword(s):

Rating Scales ◽

Double Blind ◽

Propranolol Group ◽

Schizophrenic Patients ◽

Propranolol Treatment ◽

Simultaneous Decrease ◽

The Mean ◽

Haloperidol Treatment ◽

To Receive ◽

Drug Free

Psychopharmacological interventions for managing aggression in schizophrenia have thus far yielded inconsistent results. This study evaluates the antiaggressive efficacy of combined haloperidol-propranolol treatment. Thirty-four newly admitted schizophrenic patients were studied in a controlled double-blind trial. Following a 3-day drug-free period and 7 days of haloperidol treatment, patients were randomly assigned to receive either haloperidol-propranolol or haloperidol-placebo for eight consecutive weeks. Doses of medications were adjusted as necessary; biperiden was administered if required. Rating scales were applied to assess aggression, anger, psychosis, depression, anxiety and extrapyramidal symptoms. The mean daily dose of haloperidol was 21 mg (SD = 6.4) in the research group and 29 mg (SD = 6.9) in the controls. Mean and maximal daily doses of propranolol were 159 mg (SD = 61) and 192 mg (SD = 83), and of placebo, 145 mg (SD = 50) and 180 mg (SD = 70), respectively. Compared with the controls, the scores for the research patients decreased significantly from baseline, particularly after 4 weeks of treatment, for some dimensions of anger, psychosis, anxiety, and neuroleptic-induced parkinsonism. A tendency for reduced aggression was shown in the combined haloperidol-propranolol group for some dimensions but not others. These patients also required significantly less biperiden. The tendency toward elevated antiaggressive effect of combined haloperidol-propranolol treatment compared to haloperidol alone may be explained by a simultaneous decrease in aggression, psychotic symptomatology, and anxiety.

Download Full-text

DMSA-scintigraphy in paediatrics: Is the evaluation of the geometric mean necessary for the calculation of the differential renal function?

Nuklearmedizin ◽

10.1055/s-0038-1625921 ◽

2001 ◽

Vol 40 (04) ◽

pp. 107-110 ◽

Cited By ~ 1

Author(s):

B. Roßmüller ◽

S. Alalp ◽

S. Fischer ◽

S. Dresel ◽

K. Hahn ◽

...

Keyword(s):

Renal Function ◽

Geometric Mean ◽

Region Of Interest ◽

Posterior View ◽

Renal Abnormality ◽

Anterior View ◽

Differential Renal Function ◽

The Mean ◽

The Right ◽

To Receive

SummaryFor assessment of differential renal function (PF) by means of static renal scintigraphy with Tc-99m-dimer-captosuccinic acid (DMSA) the calculation of the geometric mean of counts from the anterior and posterior view is recommended. Aim of this retrospective study was to find out, if the anterior view is necessary to receive an accurate differential renal function by calculating the geometric mean compared to calculating PF using the counts of the posterior view only. Methods: 164 DMSA-scans of 151 children (86 f, 65 m) aged 16 d to 16 a (4.7 ± 3.9 a) were reviewed. The scans were performed using a dual head gamma camera (Picker Prism 2000 XP, low energy ultra high resolution collimator, matrix 256 x 256,300 kcts/view, Zoom: 1.6-2.0). Background corrected values from both kidneys anterior and posterior were obtained. Using region of interest technique PF was calculated using the counts of the dorsal view and compared with the calculated geometric mean [SQR(Ctsdors x Ctsventr]. Results: The differential function of the right kidney was significantly less when compared to the calculation of the geometric mean (p<0.01). The mean difference between the PFgeom and the PFdors was 1.5 ± 1.4%. A difference > 5% (5.0-9.5%) was obtained in only 6/164 scans (3.7%). Three of 6 patients presented with an underestimated PFdors due to dystopic kidneys on the left side in 2 patients and on the right side in one patient. The other 3 patients with a difference >5% did not show any renal abnormality. Conclusion: The calculation of the PF from the posterior view only will give an underestimated value of the right kidney compared to the calculation of the geometric mean. This effect is not relevant for the calculation of the differntial renal function in orthotopic kidneys, so that in these cases the anterior view is not necesssary. However, geometric mean calculation to obtain reliable values for differential renal function should be applied in cases with an obvious anatomical abnormality.

Download Full-text

Implantation depth and its influence on complications after TAVI with self-expanding valves

The International Journal of Cardiovascular Imaging ◽

10.1007/s10554-021-02275-3 ◽

2021 ◽

Author(s):

Philipp Breitbart ◽

Jan Minners ◽

Manuel Hein ◽

Holger Schröfel ◽

Franz-Josef Neumann ◽

...

Keyword(s):

Characteristic Curve ◽

Fusion Imaging ◽

Paravalvular Leakage ◽

Conduction Disturbances ◽

Transcatheter Aortic Valve ◽

Transcatheter Heart Valve ◽

Implantation Depth ◽

The Mean ◽

Valve Implantation ◽

To Receive

AbstractPrior studies in patients with transcatheter aortic valve implantation (TAVI) demonstrated an influence of transcatheter heart valve (THV) position on the occurrence of new conductions disturbances (CD) and paravalvular leakage (PVL) post TAVI in balloon-expandable valves (BEV). Purpose of this study was to investigate the THV implantation depth and its influence on the occurrence of CD and PVL in self-expanding valves (SEV). We performed fusion imaging of pre- and post-procedural computed tomography angiography in 104 TAVI-patients (all with Evolut R) to receive a 3-D reconstruction of the THV within the native annulus region. The THV length below the native annulus was measured for assessment of implantation depth. Electrocardiograms pre-discharge were assessed for conduction disturbances (CD), PVL was determined in transthoracic echocardiography. The mean implantation depth of the THV in the whole cohort was 4.3 ± 3.0 mm. Using the best cut-off of ≥ 4 mm in receiver operating characteristic curve analysis (sensitivity 83.3%, specificity 60.0%) patients with lower THV position developed more new CD after TAVI (68.2 vs. 23.7%, P < 0.001). A deep THV position was identified as the only predictor for new CD after TAVI (odds ratio [CI] 1.312[1.119–1.539], P = 0.001). The implantation depth showed no influence on the grade of PVL (r = 0.052, P = 0.598). In patients with TAVI using the Evolut R SEV, a lower THV positioning (≥ 4 mm length below annulus) was a predictor for new conduction disturbances. In contrast, implantation depth was not associated with the extent of PVL. Graphic abstract Prostheses positions of self-expanding valves and their influence on the occurrence of new conduction disturbances and the grade of paravalvular leakage after TAVI.

Download Full-text

Estimation of Motion and Respiratory Characteristics during the Meditation Practice Based on Video Analysis

Sensors ◽

10.3390/s21113771 ◽

2021 ◽

Vol 21 (11) ◽

pp. 3771

Author(s):

Alexey Kashevnik ◽

Walaa Othman ◽

Igor Ryabchikov ◽

Nikolay Shilov

Keyword(s):

Respiratory Rate ◽

Evaluation System ◽

Body Movement ◽

Absolute Error ◽

The Body ◽

Meditation Practice ◽

Body Parts ◽

Mental Health Training ◽

Negative Thoughts ◽

The Mean

Meditation practice is mental health training. It helps people to reduce stress and suppress negative thoughts. In this paper, we propose a camera-based meditation evaluation system, that helps meditators to improve their performance. We rely on two main criteria to measure the focus: the breathing characteristics (respiratory rate, breathing rhythmicity and stability), and the body movement. We introduce a contactless sensor to measure the respiratory rate based on a smartphone camera by detecting the chest keypoint at each frame, using an optical flow based algorithm to calculate the displacement between frames, filtering and de-noising the chest movement signal, and calculating the number of real peaks in this signal. We also present an approach to detecting the movement of different body parts (head, thorax, shoulders, elbows, wrists, stomach and knees). We have collected a non-annotated dataset for meditation practice videos consists of ninety videos and the annotated dataset consists of eight videos. The non-annotated dataset was categorized into beginner and professional meditators and was used for the development of the algorithm and for tuning the parameters. The annotated dataset was used for evaluation and showed that human activity during meditation practice could be correctly estimated by the presented approach and that the mean absolute error for the respiratory rate is around 1.75 BPM, which can be considered tolerable for the meditation application.

Download Full-text

Towards Robust Training of Multi-Sensor Data Fusion Network Against Adversarial Examples in Semantic Segmentation

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) ◽

10.1109/icassp39728.2021.9413772 ◽

2021 ◽

Author(s):

Youngjoon Yu ◽

Hong Joo Lee ◽

Byeong Cheon Kim ◽

Jung Uk Kim ◽

Yong Man Ro

Keyword(s):

Data Fusion ◽

Semantic Segmentation ◽

Sensor Data ◽

Sensor Data Fusion ◽

Multi Sensor Data Fusion ◽

Adversarial Examples

Download Full-text

Design and Implementation of Intelligent Inspection and Alarm Flight System for Epidemic Prevention

Drones ◽

10.3390/drones5030068 ◽

2021 ◽

Vol 5 (3) ◽

pp. 68

Author(s):

Jiwei Fan ◽

Xiaogang Yang ◽

Ruitao Lu ◽

Xueli Xie ◽

Weipeng Li

Keyword(s):

Deep Learning ◽

Autonomous Navigation ◽

Detection Method ◽

Active Role ◽

Absolute Error ◽

Face Mask ◽

Learning Technology ◽

Flight System ◽

Crowd Density ◽

The Mean

Unmanned aerial vehicles (UAV) and related technologies have played an active role in the prevention and control of novel coronaviruses at home and abroad, especially in epidemic prevention, surveillance, and elimination. However, the existing UAVs have a single function, limited processing capacity, and poor interaction. To overcome these shortcomings, we designed an intelligent anti-epidemic patrol detection and warning flight system, which integrates UAV autonomous navigation, deep learning, intelligent voice, and other technologies. Based on the convolution neural network and deep learning technology, the system possesses a crowd density detection method and a face mask detection method, which can detect the position of dense crowds. Intelligent voice alarm technology was used to achieve an intelligent alarm system for abnormal situations, such as crowd-gathering areas and people without masks, and to carry out intelligent dissemination of epidemic prevention policies, which provides a powerful technical means for epidemic prevention and delaying their spread. To verify the superiority and feasibility of the system, high-precision online analysis was carried out for the crowd in the inspection area, and pedestrians’ faces were detected on the ground to identify whether they were wearing a mask. The experimental results show that the mean absolute error (MAE) of the crowd density detection was less than 8.4, and the mean average precision (mAP) of face mask detection was 61.42%. The system can provide convenient and accurate evaluation information for decision-makers and meets the requirements of real-time and accurate detection.

Download Full-text

Explanation Plus Prediction—The Logical Focus of Project Management Research

Project Management Journal ◽

10.1177/8756972821999945 ◽

2021 ◽

pp. 875697282199994

Author(s):

Joseph F. Hair ◽

Marko Sarstedt

Keyword(s):

Project Management ◽

Statistical Models ◽

Predictive Power ◽

Mean Absolute Error ◽

Explanatory Power ◽

Absolute Error ◽

Model Parameters ◽

Mean Square ◽

Management Research ◽

The Mean

Most project management research focuses almost exclusively on explanatory analyses. Evaluation of the explanatory power of statistical models is generally based on F-type statistics and the R 2 metric, followed by an assessment of the model parameters (e.g., beta coefficients) in terms of their significance, size, and direction. However, these measures are not indicative of a model’s predictive power, which is central for deriving managerial recommendations. We recommend that project management researchers routinely use additional metrics, such as the mean absolute error or the root mean square error, to accurately quantify their statistical models’ predictive power.

Download Full-text

An Estimation Method of Continuous Non-Invasive Arterial Blood Pressure Waveform Using Photoplethysmography: A U-Net Architecture-Based Approach

Sensors ◽

10.3390/s21051867 ◽

2021 ◽

Vol 21 (5) ◽

pp. 1867

Author(s):

Tasbiraha Athaya ◽

Sunwoong Choi

Keyword(s):

Blood Pressure ◽

Cardiovascular Health ◽

Estimation Method ◽

Absolute Error ◽

British Hypertension Society ◽

Non Invasive ◽

Arterial Blood ◽

Blood Pressure Waveform ◽

The Mean ◽

Mimic Iii

Blood pressure (BP) monitoring has significant importance in the treatment of hypertension and different cardiovascular health diseases. As photoplethysmogram (PPG) signals can be recorded non-invasively, research has been highly conducted to measure BP using PPG recently. In this paper, we propose a U-net deep learning architecture that uses fingertip PPG signal as input to estimate arterial BP (ABP) waveform non-invasively. From this waveform, we have also measured systolic BP (SBP), diastolic BP (DBP), and mean arterial pressure (MAP). The proposed method was evaluated on a subset of 100 subjects from two publicly available databases: MIMIC and MIMIC-III. The predicted ABP waveforms correlated highly with the reference waveforms and we have obtained an average Pearson’s correlation coefficient of 0.993. The mean absolute error is 3.68 ± 4.42 mmHg for SBP, 1.97 ± 2.92 mmHg for DBP, and 2.17 ± 3.06 mmHg for MAP which satisfy the requirements of the Association for the Advancement of Medical Instrumentation (AAMI) standard and obtain grade A according to the British Hypertension Society (BHS) standard. The results show that the proposed method is an efficient process to estimate ABP waveform directly using fingertip PPG.

Download Full-text

A Multi-Channel Uncertainty-Aware Multi-Resolution Network for MR to CT Synthesis

Applied Sciences ◽

10.3390/app11041667 ◽

2021 ◽

Vol 11 (4) ◽

pp. 1667

Author(s):

Kerstin Klaser ◽

Pedro Borges ◽

Richard Shaw ◽

Marta Ranzini ◽

Marc Modat ◽

...

Keyword(s):

Medical Image Analysis ◽

Absolute Error ◽

Magnetic Resonance Images ◽

The Body ◽

Whole Body ◽

Image Size ◽

Body Images ◽

Image Translation ◽

The Mean ◽

Synthetic Ct

Synthesising computed tomography (CT) images from magnetic resonance images (MRI) plays an important role in the field of medical image analysis, both for quantification and diagnostic purposes. Convolutional neural networks (CNNs) have achieved state-of-the-art results in image-to-image translation for brain applications. However, synthesising whole-body images remains largely uncharted territory, involving many challenges, including large image size and limited field of view, complex spatial context, and anatomical differences between images acquired at different times. We propose the use of an uncertainty-aware multi-channel multi-resolution 3D cascade network specifically aiming for whole-body MR to CT synthesis. The Mean Absolute Error on the synthetic CT generated with the MultiResunc network (73.90 HU) is compared to multiple baseline CNNs like 3D U-Net (92.89 HU), HighRes3DNet (89.05 HU) and deep boosted regression (77.58 HU) and shows superior synthesis performance. We ultimately exploit the extrapolation properties of the MultiRes networks on sub-regions of the body.

Download Full-text