scholarly journals Improving Performance of the PRYSTINE Traffic Sign Classification by Using Perturbation-based Explainability Approach

2021 ◽  
Author(s):  
Ivars Namatēvs ◽  
Kaspars Sudars ◽  
Kaspars Ozols
Keyword(s):  
Autonomous Driving ◽  
Traffic Light ◽  
Traffic Sign ◽  
High Stakes ◽  
Box Models ◽  
Classifier Performance ◽  
The Status ◽  
Post Hoc ◽  
Fully Connected ◽  
Kernel Compression

Model understanding is critical in many domains, particularly those involved in high-stakes decisions, i.e., medicine, criminal justice, and autonomous driving. Explainable AI (XAI) methods are essential for working with black-box models such as Convolutional Neural Networks. This paper evaluates the traffic sign classifier of Deep Neural Network (DNN) from the Programmable Systems for Intelligence in Automobiles (PRYSTINE) project for explainability. The results of explanations were further used for the CNN PRYSTINE classifier vague kernels` compression. After all, the precision of the classifier was evaluated in different pruning scenarios. The proposed classifier performance methodology was realised by creating the original traffic sign and traffic light classification and explanation code. First, the status of the kernels of the network was evaluated for explainability. For this task, the post-hoc, local, meaningful perturbation-based forward explainable method was integrated into the model to evaluate each kernel status of the network. This method enabled distinguishing high and low-impact kernels in the CNN. Second, the vague kernels of the classifier of the last layer before the fully connected layer were excluded by withdrawing them from the network. Third, the network's precision was evaluated in different kernel compression levels. It is shown that by using the XAI approach for network kernel compression, the pruning of 5% of kernels leads only to a 1% loss in traffic sign and traffic light classification precision. The proposed methodology is crucial where execution time and processing capacity prevail.

The Traffic Sign Detection Algorithm Based on Region of Interest Extraction and Double Filter

2019 ◽  
Vol 13 ◽  
Author(s):  
Dongxian Yu ◽  
Jiatao Kang ◽  
Zaihui Cao ◽  
Neha Jain
Keyword(s):  
Region Of Interest ◽  
Detection Algorithm ◽  
Light Filter ◽  
Support Vector ◽  
Detection Accuracy ◽  
Traffic Light ◽  
Traffic Sign ◽  
Traffic Signs ◽  
Sign Detection ◽  
Traffic Sign Detection

In order to solve the current traffic sign detection technology due to the interference of various complex factors, it is difficult to effectively carry out the correct detection of traffic signs, and the robustness is weak, a traffic sign detection algorithm based on the region of interest extraction and double filter is designed.First, in order to reduce environmental interference, the input image is preprocessed to enhance the main color of each logo.Secondly, in order to improve the extraction ability Of Regions Of Interest, a Region Of Interest (ROI) detector based on Maximally Stable Extremal Regions (MSER) and Wave Equation (WE) was defined, and candidate Regions were selected through the ROI detector.Then, an effective HOG (Histogram of Oriented Gradient) descriptor is introduced as the detection feature of traffic signs, and SVM (Support Vector Machine) is used to classify them into traffic signs or background.Finally, the context-aware filter and the traffic light filter are used to further identify the false traffic signs and improve the detection accuracy.In the GTSDB database, three kinds of traffic signs, which are indicative, prohibited and dangerous, are tested, and the results show that the proposed algorithm has higher detection accuracy and robustness compared with the current traffic sign recognition technology.

scholarly journals Democratizing ownership and participation in the 4th Industrial Revolution: challenges and opportunities in cellular agriculture

2021 ◽  
Author(s):  
Robert M. Chiles ◽  
Garrett Broad ◽  
Mark Gagnon ◽  
Nicole Negowetti ◽  
Leland Glenna ◽  
...  
Keyword(s):  
Business Models ◽  
Industrial Revolution ◽  
Cultured Cells ◽  
Advanced Materials ◽  
High Stakes ◽  
Digital Platforms ◽  
Challenges And Opportunities ◽  
The Status ◽  
Political Economic ◽  
Industry Sectors

AbstractThe emergence of the “4th Industrial Revolution,” i.e. the convergence of artificial intelligence, the Internet of Things, advanced materials, and bioengineering technologies, could accelerate socioeconomic insecurities and anxieties or provide beneficial alternatives to the status quo. In the post-Covid-19 era, the entities that are best positioned to capitalize on these innovations are large firms, which use digital platforms and big data to orchestrate vast ecosystems of users and extract market share across industry sectors. Nonetheless, these technologies also have the potential to democratize ownership, broaden political-economic participation, and reduce environmental harms. We articulate the potential sociotechnical pathways in this high-stakes crossroads by analyzing cellular agriculture, an exemplary 4th Industrial Revolution technology that synergizes computer science, biopharma, tissue engineering, and food science to grow cultured meat, dairy, and egg products from cultured cells and/or genetically modified yeast. Our exploration of this space involved multi-sited ethnographic research in both (a) the cellular agriculture community and (b) alternative economic organizations devoted to open source licensing, member-owned cooperatives, social financing, and platform business models. Upon discussing how these latter approaches could potentially facilitate alternative sociotechnical pathways in cellular agriculture, we reflect upon the broader implications of this work with respect to the 4th Industrial Revolution and the enduring need for public policy reform.

scholarly journals Post hoc analysis of ellipsoid zone changes beyond the central subfield in symptomatic vitreomacular adhesion patients from the OASIS trial

2021 ◽  
Vol 6 (1) ◽  
pp. e000648
Author(s):  
Swetha Bindu Velaga ◽  
Muneeswar Gupta Nittala ◽  
Michael S Ip ◽  
Luc Duchateau ◽  
SriniVas R Sadda
Keyword(s):  
Time Course ◽  
Vitreomacular Adhesion ◽  
Ellipsoid Zone ◽  
Post Hoc Analysis ◽  
Link Type ◽  
Phase Iiib ◽  
The Status ◽  
Registration Number ◽  
Post Hoc ◽  
And Control

Background/aimsOASIS is a Phase IIIb trial (NCT01429441) assessing long-term outcomes in subjects with symptomatic vitreomacular adhesion (VMA). The purpose of this study is to report on the frequency, severity, location and time course of ellipsoid zone (EZ) alterations in ocriplasmin-treated and sham control eyes in the OASIS study.Methods220 patients (146 ocriplasmin, 74 sham) subjects with VMA were enrolled in this masked post hoc analysis phase IIIb, randomised, sham-controlled double-masked multicentre clinical trial. A masked post hoc analysis of OCT images was performed at the Doheny Image Reading Center from subjects enrolled in the OASIS trial. The status of the EZ band was assessed in three different macular regions: the central subfield (CS) (≤1 mm diameter), the parafoveal area (PAA) (>1 to ≤3 mm) and the perifoveal area (PEA) (>3 to ≤6 mm). The EZ band was rated as normal/intact, full thickness macular hole (FTMH), abnormal but continuous, discontinuous/disrupted or absent at visits from baseline (pretreatment) to week 1 (day 7), month 1 (day 28), month 3, month 6, month 12 and the final follow-up at month 24. EZ band status was compared in both study and control eyes.ResultsA total of 208 patients (138 ocriplasmin, 70 sham) were included in this analysis. At baseline, FTMH was present in 48.6%, 8.0%, 0% and 52.8%, 2.9%, 0% in the CS, PAA and PEA of the ocriplasmin and sham groups, respectively. The EZ was graded to be abnormal but continuous, discontinuous/disrupted or absent at Baseline in 21.0%, 4.3%, 2.8% in the CS, PAA and PEA, respectively, of the ocriplasmin group; and 12.9%, 10.0%, 4.3% in the CS, PAA and PEA of the sham group. For the ocriplasmin group in the PAA, this frequency increased to 6.6% at week 1, was 9.8% at month 1, but improved to 3.8% at month 3, and remained stable to 1.6% at month 24. These differences, however, were not statistically significant.ConclusionsOcriplasmin treatment for symptomatic VMA was associated with EZ abnormalities in a small percentage of patients that was best assessed in regions (PEA) relatively unaffected by the VM interface disease at baseline. The EZ abnormalities were apparent by week 1, persisted at month 1, and appeared to resolve in the majority of cases by month 3.Trial registration numberNCT01429441

scholarly journals The natural duty of justice in non-ideal circumstances: On the moral demands of institution building and reform

2017 ◽  
Vol 20 (1) ◽  
pp. 45-66 ◽  
Author(s):  
Laura Valentini
Keyword(s):  
Distributive Justice ◽  
Political Theory ◽  
Status Quo ◽  
The State ◽  
Response Strategy ◽  
Macro Level ◽  
Institution Building ◽  
Institutional Agents ◽  
High Stakes ◽  
The Status

Principles of distributive justice bind macro-level institutional agents, like the state. But what does justice require in non-ideal circumstances, where institutional agents are unjust or do not exist in the first place? Many answer by invoking Rawls's natural duty ‘to further just arrangements not yet established’, treating it as a ‘normative bridge’ between institutional demands of distributive justice and individual responsibilities in non-ideal circumstances. I argue that this response strategy is unsuccessful. I show that the more unjust the status quo is due to non-compliance, the less demanding the natural duty of justice becomes. I conclude that, in non-ideal circumstances, the bulk of the normative work is done by another natural duty: that of beneficence. This conclusion has significant implications for how we conceptualize our political responsibilities in non-ideal circumstances, and cautions us against the tendency – common in contemporary political theory – to answer all high-stakes normative questions under the rubric of justice.

scholarly journals Aplicación móvil para el seguimiento de solicitudes de acceso a la información

2020 ◽  
pp. 21-29
Author(s):  
Eva Rafael-Pérez ◽  
Eliel López-Cruz ◽  
Alan Jhaseel Hernández-Bolaños ◽  
Bibiana Díaz-Sarmiento
Keyword(s):  
Internal Control ◽  
Mobile Application ◽  
Public Information ◽  
The State ◽  
Extreme Programming ◽  
Access To Information ◽  
Traffic Light ◽  
Development Methodology ◽  
The Status ◽  
The Times

Currently, the use of mobile applications is of utmost importance, since it allows the availability of information at all times. The mobile application described in this article was made to follow up on the Requests for Access to Information and is part of a web system, with which an internal control of the Requests for Access to Information that are sent to the Dependencies, Entities, Auxiliary Bodies and Trusts of the State Public Administration. The application will attend to the requests and thus, the times of attention to them are met within the terms established by the Law of Transparency and Access to Public Information for the State of Oaxaca (LGTAIPO). The mobile application allows the monitoring of information requests received, shows the status in which they are through a colored traffic light, meetings can be scheduled, allows the registration of the user's profile, and, has an internal chat so that dependencies can communicate with each other and send notifications. The development methodology used for this project was Extreme Programming using the Dart programming language.

A collection of easily deployable adversarial traffic sign stickers

2021 ◽  
Vol 69 (6) ◽  
pp. 511-523
Author(s):  
Henrietta Lengyel ◽  
Viktor Remeli ◽  
Zsolt Szalay
Keyword(s):  
Traffic Safety ◽  
Autonomous Driving ◽  
Driver Assistance ◽  
Driver Assistance Systems ◽  
Traffic Sign ◽  
Practical Possibility ◽  
Special Knowledge ◽  
Environment Perception ◽  
Assistance Systems ◽  
Traffic Sign Classification

Abstract The emergence of new autonomous driving systems and functions – in particular, systems that base their decisions on the output of machine learning subsystems responsible for environment perception – brings a significant change in the risks to the safety and security of transportation. These kinds of Advanced Driver Assistance Systems are vulnerable to new types of malicious attacks, and their properties are often not well understood. This paper demonstrates the theoretical and practical possibility of deliberate physical adversarial attacks against deep learning perception systems in general, with a focus on safety-critical driver assistance applications such as traffic sign classification in particular. Our newly developed traffic sign stickers are different from other similar methods insofar that they require no special knowledge or precision in their creation and deployment, thus they present a realistic and severe threat to traffic safety and security. In this paper we preemptively point out the dangers and easily exploitable weaknesses that current and future systems are bound to face.

scholarly journals Preliminary establishment and analysis of automated driving benchmark in China

2019 ◽  
Vol 259 ◽  
pp. 02002
Author(s):  
Xiong Hui ◽  
Yinghan Wang ◽  
Qiang Ge ◽  
Ziqing Gu ◽  
Mingyang Cui ◽  
...  
Keyword(s):  
Large Scale ◽  
Road User ◽  
Future Application ◽  
Area Network ◽  
Traffic Light ◽  
Automated Driving ◽  
Traffic Sign ◽  
Traffic Characteristics ◽  
Chinese Characteristic ◽  
And Control

In order to promote the localization of Automated Driving (AD) in China, it is necessary to collect large-scale traffic scene data with Chinese characteristic for future analysis. In this paper, we propose the methodologies and rules of establishing AD benchmark involving how to configure sensors, how to design the collection schema to show Chinese traffic characteristics and the rules of elaborating distinctive scenes and routes, what to label, and it is also demonstrated that the benchmark can support the future application of extended AD research. Data collection lasted about one month covering diverse scene data such as campus, highway, park, etc. from three representative Chinese cities and driving data from 30 different drivers. Moreover, some statistical results and analyses are produced in accordance with the designed methodologies as instances of potential application. Up to now, the dataset contains about 7,000 labelled image frames and corresponding LiDAR, GPS and Controller Area Network (CAN) data. Labels cover scene type, road user, traffic sign, traffic light, and lane marker. This benchmark can help researchers better understand Chinese traffic situation in aspects of environmental perception, driving behavior analysis, risk assessment, automated vehicle decision and control.

Winning Together

2018 ◽  
Author(s):  
Bruno Verdini Trejo
Keyword(s):  
Resource Management ◽  
Energy Transition ◽  
The United States ◽  
Severe Drought ◽  
High Stakes ◽  
Climate Risks ◽  
Environmental Disputes ◽  
First Case ◽  
The Status ◽  
Oil And Natural Gas

Through an analysis of prominent transboundary natural resource management negotiation cases, Winning Together outlines how government, industry, and NGOs can effectively overcome past grievances, break the status quo, resolve conflicts, and create mutual gains in high-stakes water, energy, and environmental disputes. The book examines two landmark international negotiations between the United States and Mexico, both with agreements signed in 2012 after several decades of deadlock. The first case involves the conflict over the shared hydrocarbon reservoirs in the Gulf of Mexico, containing significant oil and natural gas resources. The second analyzes the dispute, amidst severe drought and increased climate risks, over the environmental resources and shared waters of the Colorado River, providing irrigation and water supply to more than 40 million people. For the first time, the two countries established a binational framework to co-develop and jointly manage these transboundary natural resources, as partners. Through unprecedented interviews with over 70 negotiators on both sides of the border, the book underscores strategies by which resource management practitioners can effectively increase river basin supply, re-think irrigation and storage infrastructure, restore ecosystems and habitats, enhance coordination between private and state owned companies, improve energy transition and planning, and re-define the scope and impact of diplomatic partnerships. Winning Together shows how developed and developing countries can move beyond hard-bargaining tactics and avoid the ultimatums that accompany the presumption that there are not enough resources to go around, and that one side must win and the other must inevitably lose.

Sign in / Sign up

Export Citation Format

Share Document