Predicting Near-Term Train Schedule Performance and Delay Using Bi-Level Random Forests

2019 ◽  
Vol 2673 (5) ◽  
pp. 564-573 ◽  
Author(s):  
Mohammad Amin Nabian ◽  
Negin Alemazkoor ◽  
Hadi Meidani
Keyword(s):  
Time Frame ◽  
Learning Approaches ◽  
Arrival Times ◽  
Primary Level ◽  
Passenger Train ◽  
Proposed Model ◽  
Level Model ◽  
Schedule Performance ◽  
Near Term ◽  
Delay Prediction

Accurate near-term passenger train delay prediction is critical for optimal railway management and providing passengers with accurate train arrival times. In this work, a novel bi-level random forest approach is proposed to predict passenger train delays in the Netherlands. The primary level predicts whether a train delay will increase, decrease, or remain unchanged in a specified time frame. The secondary level then estimates the actual delay (in minutes), given the predicted delay category at primary level. For validation purposes, the proposed model has been compared with several alternative statistical and machine-learning approaches. The results show that the proposed model provides the best prediction accuracy compared with other alternatives. Moreover, constructing the proposed bi-level model is computationally cheap, thereby being easily applicable.

scholarly journals Deep learning systems detect dysplasia with human-like accuracy using histopathology and probe-based confocal laser endomicroscopy

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Shan Guleria ◽  
Tilak U. Shah ◽  
J. Vincent Pulido ◽  
Matthew Fasullo ◽  
Lubaina Ehsan ◽  
...  
Keyword(s):  
Deep Learning ◽  
Diagnostic Accuracy ◽  
High Sensitivity ◽  
Confocal Laser Endomicroscopy ◽  
Confocal Laser ◽  
Learning Approaches ◽  
Learning Models ◽  
Whole Slide Image ◽  
Slide Image ◽  
Level Model

AbstractProbe-based confocal laser endomicroscopy (pCLE) allows for real-time diagnosis of dysplasia and cancer in Barrett’s esophagus (BE) but is limited by low sensitivity. Even the gold standard of histopathology is hindered by poor agreement between pathologists. We deployed deep-learning-based image and video analysis in order to improve diagnostic accuracy of pCLE videos and biopsy images. Blinded experts categorized biopsies and pCLE videos as squamous, non-dysplastic BE, or dysplasia/cancer, and deep learning models were trained to classify the data into these three categories. Biopsy classification was conducted using two distinct approaches—a patch-level model and a whole-slide-image-level model. Gradient-weighted class activation maps (Grad-CAMs) were extracted from pCLE and biopsy models in order to determine tissue structures deemed relevant by the models. 1970 pCLE videos, 897,931 biopsy patches, and 387 whole-slide images were used to train, test, and validate the models. In pCLE analysis, models achieved a high sensitivity for dysplasia (71%) and an overall accuracy of 90% for all classes. For biopsies at the patch level, the model achieved a sensitivity of 72% for dysplasia and an overall accuracy of 90%. The whole-slide-image-level model achieved a sensitivity of 90% for dysplasia and 94% overall accuracy. Grad-CAMs for all models showed activation in medically relevant tissue regions. Our deep learning models achieved high diagnostic accuracy for both pCLE-based and histopathologic diagnosis of esophageal dysplasia and its precursors, similar to human accuracy in prior studies. These machine learning approaches may improve accuracy and efficiency of current screening protocols.

scholarly journals Lane Allocation Optimization in Container Seaport Gate System Considering Carbon Emissions

2021 ◽  
Vol 13 (7) ◽  
pp. 3628
Author(s):  
Zhihong Jin ◽  
Xin Lin ◽  
Linlin Zang ◽  
Weiwei Liu ◽  
Xisheng Xiao
Keyword(s):  
Branch And Bound ◽  
Carbon Emissions ◽  
Integer Program ◽  
Arrival Times ◽  
Proposed Model ◽  
Gate System ◽  
Appointment System ◽  
Truck Appointment System ◽  
Gate Congestion ◽  
Allocation Decisions

Long queues of arrival trucks are a common problem in seaports, and thus, carbon emissions generated from trucks in the queue cause environmental pollution. In order to relieve gate congestion and reduce carbon emissions, this paper proposes a lane allocation framework combining the truck appointment system (TAS) for four types of trucks. Based on the distribution of arrival times obtained from the TAS, lane allocation decisions in each appointment period are determined in order to minimize the total cost, including the operation cost and carbon emissions cost. The resultant optimization model is a non-linear fractional integer program. This model was firstly transformed to an equivalent integer program with bilinear constraints. Then, an improved branch-and-bound algorithm was designed, which includes further transforming the program into a linear program using the McCormick approximation method and iteratively generating a tighter outer approximation along the branch-and-bound procedure. Numerical studies confirmed the validity of the proposed model and algorithm, while demonstrating that the lane allocation decisions could significantly reduce carbon emissions and operation costs.

Spatio-Temporal Gaussianization Flows for Extreme Event Detection

2021 ◽  
Author(s):  
Miguel-Ángel Fernández-Torres ◽  
J. Emmanuel Johnson ◽  
María Piles ◽  
Gustau Camps-Valls
Keyword(s):  
Event Detection ◽  
Land Surface ◽  
Extreme Event ◽  
Learning Approaches ◽  
Proposed Model ◽  
Individual Contributions ◽  
Current Context ◽  
Spatio Temporal ◽  
The Individual ◽  
Anomaly Score

<p>Automatic anticipation and detection of extreme events constitute a major challenge in the current context of climate change. Machine learning approaches have excelled in detection of extremes and anomalies in Earth data cubes recently, but are typically both computationally costly and supervised, which hamper their wide adoption. We alternatively present here an unsupervised, efficient, generative approach for extreme event detection, whose performance is illustrated for drought detection in Europe during the severe Russian heat wave in 2010. The core architecture of the model is generic and could naturally be extended to the detection of other kinds of anomalies. First, it computes hierarchical appearance (spatial) and motion (temporal) representations of several informative Essential Climate Variables (ECVs), including soil moisture, land surface temperature, as well as features describing vegetation health. Then, these representations are combined using Gaussianization Flows that yield a spatio-temporal anomaly score. This allows the proposed model not only to detect droughts areas, but also to explain why they were produced, monitoring the individual contributions of each of the ECVs to the indicator at its output.</p>

Fast Transaction-Level Model for Direct Memory Access Controller

2019 ◽  
Vol 28 (04) ◽  
pp. 1950059
Author(s):  
Mona Safar ◽  
Magdy A. El-Moursy ◽  
Ahmed Tarek ◽  
Ahmed Emad ◽  
Ahmed Hesham ◽  
...  
Keyword(s):  
Open Source ◽  
Direct Memory Access ◽  
System Level ◽  
The Past ◽  
Simulation Speed ◽  
Proposed Model ◽  
Level Model ◽  
Data Bus ◽  
Transaction Level ◽  
Simulation Time

Transaction-Level Modeling (TLM) has been widely used in system-level design in the past few years. Simulation speed of Virtual Platforms (VPs) depends mainly on the transactions which are initiated by the Programmer’s View (PV) models of the VP devices. PV models are required to run at highest simulation speed. Data bus width as a hardware (HW) parameter should not reduce simulation speed of the modeled transactions. Furthermore, HW-related parameters should only be accounted for when considering timing of the models. A fast SystemC-TLM model is developed for the widely used ARM PrimeCell PL080 DMAC IP. The performance of the proposed model is validated against a developed RTL model for the same device. The effect of the transactions granularity on simulation speed is determined. Different programmed transfers are simulated and compared with open-source Quick Emulator (QEMU)-based models. The developed model is compared with the developed RTL, the open-source QEMU model, and the existing ARM Fast Model (AFM). It is shown that simulation time of the developed model is reduced by two orders of magnitude as compared to the other existing models.

Feeder Bus Timetable Design and Vehicle Size Setting in Peak Hour Demand Conditions

2019 ◽  
Vol 2673 (10) ◽  
pp. 321-332 ◽  
Author(s):  
Xueping Dou ◽  
Qiang Meng
Keyword(s):  
Mixed Integer ◽  
Arrival Times ◽  
Non Linear Programming ◽  
Bee Colony ◽  
Proposed Model ◽  
Feeder Bus ◽  
Minlp Model ◽  
Failure Cost ◽  
Set Up ◽  
The Given

This study proposes a solution to the feeder bus timetabling problem, in which the terminal departure times and vehicle sizes are simultaneously determined based on the given transfer passengers and their arrival times at a bus terminal. The problem is formulated as a mixed integer non-linear programming (MINLP) model with the objective of minimizing the transfer waiting time of served passengers, the transfer failure cost of non-served passengers, and the operating costs of bus companies. In addition to train passengers who plan to transfer to buses, local passengers who intend to board buses are considered and treated as passengers from virtual trains in the proposed model. Passenger attitudes and behaviors toward the waiting queue caused by bus capacity constraints in peak hour demand conditions are explicitly embedded in the MINLP model. A hybrid artificial bee colony (ABC) algorithm is developed to solve the MINLP model. Various experiments are set up to account for the performance of the proposed model and solution algorithm.

scholarly journals What is Quantum Computing and How it Works

2020 ◽  
pp. 1-5
Author(s):  
Bahman Zohuri ◽  
◽  
Farhang Mossavar Rahmani ◽  
Keyword(s):  
Artificial Intelligence ◽  
Quantum Computing ◽  
Quantum Computer ◽  
Time Frame ◽  
Basic Unit ◽  
Next Generation ◽  
Intel Corporation ◽  
Encrypt Data ◽  
Near Term ◽  
Incremental Step

Companies such as Intel as a pioneer in chip design for computing are pushing the edge of computing from its present Classical Computing generation to the next generation of Quantum Computing. Along the side of Intel corporation, companies such as IBM, Microsoft, and Google are also playing in this domain. The race is on to build the world’s first meaningful quantum computer—one that can deliver the technology’s long-promised ability to help scientists do things like develop miraculous new materials, encrypt data with near-perfect security and accurately predict how Earth’s climate will change. Such a machine is likely more than a decade away, but IBM, Microsoft, Google, Intel, and other tech heavyweights breathlessly tout each tiny, incremental step along the way. Most of these milestones involve packing more quantum bits, or qubits—the basic unit of information in a quantum computer—onto a processor chip ever. But the path to quantum computing involves far more than wrangling subatomic particles. Such computing capabilities are opening a new area into dealing with the massive sheer volume of structured and unstructured data in the form of Big Data, is an excellent augmentation to Artificial Intelligence (AI) and would allow it to thrive to its next generation of Super Artificial Intelligence (SAI) in the near-term time frame.

scholarly journals A Python tool and database of Amtrak departure and arrival times with weather information

2021 ◽  
Author(s):  
Pipatphon Lapamonpinyo ◽  
Sybil Derrible ◽  
Francesco Corman
Keyword(s):  
United States ◽  
The United States ◽  
Operating Characteristics ◽  
Raw Data ◽  
Arrival Times ◽  
Passenger Train ◽  
Weather Information ◽  
Weather Underground ◽  
The One ◽  
Rail Station

This article proposes a Python-based Amtrak and Weather Underground (PAWU) tool to collect data on Amtrak (the main passenger train operator in the United States) departure and arrival times with weather information. In addition, this article offers a database, developed with PAWU, of the operating characteristics of 16 Amtrak routes from 2008 to 2019. More specifically, PAWU enables users to retrieve Amtrak departure and arrival times of any train number throughout the United States. It then automatically retrieves weather information from Weather Underground for each rail station and stores the data collected in a local MySQL database. Users can easily select any desired train number(s) and date range(s) without dealing with the code and the raw data from those sources that are in different formats. The database itself can be used, in part, to develop, apply, and benchmark models that assess the performance of rail services such as the one offered by Amtrak.

Abstract P257: Getting to the Golden Hour: Correlating “Door-To” Times with 60-Minute Administration of tPA in an Academic Medical Center

2011 ◽  
Vol 4 (suppl_2) ◽  
Author(s):  
Ann M Leonhardt ◽  
Curtis G Benesch ◽  
Kate C Young
Keyword(s):  
Medical Center ◽  
Academic Medical Center ◽  
Rank Correlation ◽  
Correlation Coefficients ◽  
Time Frame ◽  
Arrival Times ◽  
Golden Hour ◽  
Stroke Team ◽  
Time Standards ◽  
Scheduling Protocols

Introduction: The efficacy of intravenous tPA for the treatment of acute stroke diminishes over time. The AHA/ASA and NINDS recommend a goal door to needle time of 60 minutes or less. Objective: Identify potential barriers to tPA administration within 60 minutes of arrival. Methods: Retrospective review of tPA adsinistration using “Get With the Guidelines” (GWTG) and institutional records from January 1, 2009 through December 31, 2010 (n=100). Spearman rank correlation coefficients were calculated for the NINDS recommended time standards, age and NIH Stroke Scale (NIHSS) score. We used a receiver-operator curve (ROC) to identify the door to CT time predictive of tPA administration ≤ 60 minutes. Results: Median door to physician, door to CT, and door to stroke team times were within the recommended goals. Door to CT (ρ=0.53, p<0.0001), and door to stroke team (ρ=0.33, p<0.01) times were positively correlated with door to tPA times. Last known well to arrival (ρ= -0.28, p<0.01) and NIHSS (ρ= -0.32, p<0.01) were negatively correlated with door to tPA times; patients with higher NIHSS and longer last-known-well to arrival times received tPA in a shorter time frame. Age and door to physician time were not correlated with tPA treatment times. After adjusting for the other benchmarks and NIHSS, only door to CT remained significantly correlated with door to IV tPA (partial correlation coefficient=0.40, p<0.001). The ROC curve showed that a goal time of 20 minutes or less for door to CT initiation had the best sensitivity and specificity for predicting tPA administration within 60 minutes. Conclusion: In keeping with the recommended time goals, median times for the intermediate steps were within target. Our median tPA times, however, did not meet the 60 minute goal. Door to CT initiation was the variable that most strongly correlated with door to needle times. Process issues such as order entry and scheduling protocols may be barriers to obtaining CT within the 20 minute time frame identified by our analysis. Other barriers after the CT scan is obtained must be identified to facilitate faster tPA administration. Further evaluation of these factors is warranted to better ensure the timely delivery of tPA to stroke patients, thereby improving patient outcomes.

scholarly journals Multi-Person Pose Estimation using an Orientation and Occlusion Aware Deep Learning Network

Sensors ◽  
2020 ◽  
Vol 20 (6) ◽  
pp. 1593 ◽  
Author(s):  
Yanlei Gu ◽  
Huiyang Zhang ◽  
Shunsuke Kamijo
Keyword(s):  
Deep Learning ◽  
Pose Estimation ◽  
Body Orientation ◽  
Learning Approaches ◽  
The Public ◽  
Learning Network ◽  
Proposed Model ◽  
Body Boundary ◽  
Public Dataset ◽  
Deep Learning Network

Image based human behavior and activity understanding has been a hot topic in the field of computer vision and multimedia. As an important part, skeleton estimation, which is also called pose estimation, has attracted lots of interests. For pose estimation, most of the deep learning approaches mainly focus on the joint feature. However, the joint feature is not sufficient, especially when the image includes multi-person and the pose is occluded or not fully visible. This paper proposes a novel multi-task framework for the multi-person pose estimation. The proposed framework is developed based on Mask Region-based Convolutional Neural Networks (R-CNN) and extended to integrate the joint feature, body boundary, body orientation and occlusion condition together. In order to further improve the performance of the multi-person pose estimation, this paper proposes to organize the different information in serial multi-task models instead of the widely used parallel multi-task network. The proposed models are trained on the public dataset Common Objects in Context (COCO), which is further augmented by ground truths of body orientation and mutual-occlusion mask. Experiments demonstrate the performance of the proposed method for multi-person pose estimation and body orientation estimation. The proposed method can detect 84.6% of the Percentage of Correct Keypoints (PCK) and has an 83.7% Correct Detection Rate (CDR). Comparisons further illustrate the proposed model can reduce the over-detection compared with other methods.

scholarly journals Supervised Learning-Based Fast, Stealthy, and Active NAT Device Identification Using Port Response Patterns

Symmetry ◽  
2020 ◽  
Vol 12 (9) ◽  
pp. 1444
Author(s):  
Seungwoon Lee ◽  
Si Jung Kim ◽  
Jungtae Lee ◽  
Byeong-hee Roh
Keyword(s):  
Supervised Learning ◽  
Response Patterns ◽  
Learning Approaches ◽  
Security Issue ◽  
Ip Address ◽  
Device Identification ◽  
Address Translation ◽  
Proposed Model ◽  
Network Operations ◽  
Network Administrators

Although network address translation (NAT) provides various advantages, it may cause potential threats to network operations. For network administrators to operate networks effectively and securely, it may be necessary to verify whether an assigned IP address is using NAT or not. In this paper, we propose a supervised learning-based active NAT device (NATD) identification using port response patterns. The proposed model utilizes the asymmetric port response patterns between NATD and non-NATD. In addition, to reduce the time and to solve the security issue that supervised learning approaches exhibit, we propose a fast and stealthy NATD identification method. The proposed method can perform the identification remotely, unlike conventional methods that should operate in the same network as the targets. The experimental results demonstrate that the proposed method is effective, exhibiting a F1 score of over 90%. With the efficient features of the proposed methods, we recommend some practical use cases that can contribute to managing networks securely and effectively.

Sign in / Sign up

Export Citation Format

Share Document