Boston AI Institute

Pay After Placement

ML Applications in Transportation Engineering - DEADLINE 31 August 2022
Transportation systems are complex, diverse, and dynamic in nature and operation. Researchers and practitioners have recently been faced with difficulties in obtaining comprehensive and current data needed to tackle rapidly emerging challenges, such as congestion of infrastructures, safety problems, environmental impacts, energy dependency, and social equity. With the rapid digitization and implementation of sensors in transport systems (e.g., personal devices, vehicles, infrastructures—including streets and sidewalks at the urban scale), there is a substantial wealth of data related to transport complex phenomena. Due to their advanced computation and data collection processes, machine learning is a fast and powerful tool that breaks down such complex problems into more straightforward and manageable mathematical operations. Researchers have developed machine learning methods to approach more traditional and novel transportation research problems with varying levels of success.

Machine learning encompasses many methodologies (e.g., supervised learning, unsupervised learning, reinforcement learning, and self-supervised learning, among others) and models (e.g., deep learning, support vector machines, decision trees, and evolutionary algorithms, among others) to explore new data sources and applications. Besides their improved performance compared to more conventional methods, machine learning could evolve to support planning and policy making in the transport field and, therefore, achieve more interpretable models and results (i.e., explainable artificial intelligence).

This Special Issue aims to collect and report new and innovative applications of machine learning methods to solve challenges presented by transportation systems. The scope of the research is diverse; topics of interest include, but are not limited to, the application of machine learning in various transportation fields and the following topics:

- Safety of transport infrastructures, particularly road users and vulnerable road users (pedestrians, cyclists, and scooter users);

- Monitoring, operation control, and management of mobility services, including shared-mobility services, public transportation management, Mobility-as-a-Service (MaaS), etc.;

- Intelligent transportation systems;

- Smart city logistics and micro-logistics;

- Management of public space management at the urban scale, including the intermittent and dynamic usage of road carriageways;

- Case studies in which machine learning was effectively used to make transportation systems more effective;

- Comparison of different approaches of machine learning methods with conventional approaches;

We welcome both original research and review articles. All submissions will be peer-reviewed according to the high standards of the journal.

Dr. Filipe Moura
Dr. Manuel Marques
Guest Editors

Deadline for manuscript submissions: 31 August 2022.

More info: https://www.mdpi.com/journal/applsci/special_issues/Machine_Learning_Applications_Transportation_Engineering

It is important to take care of your mental health.
If you’re experiencing some of these symptoms:
❗️Headache
❗️Neck & shoulder pain
❗️Upset stomach
❗️Heavy chest
maybe you’re stressed. Here is some advice that can help you to cope with stress: https://bit.ly/WHOStressManagement

A research grant is available at the University of Salento, Lecce, Italy.
The research shall be developed in close collaboration with Echolight (https://www.echolightmedical.com, 12 months out of 18 in total), and is related to the following topic:

The main goal of the project is to improve the tuning and calibration process of noninvasive diagnostic imaging devices used for imaging. One of the most critical steps during the implementation of a diagnostic imaging device is its calibration. In fact, poor calibration can lead to unreliable instrument performance with noisy images and the presence of unwanted artefacts that could mislead the diagnosis made by the physician. The calibration phase involves a repeated try-and-check procedure during which the instrument parameters are repeatedly changed in order to obtain images that are sharp and as closely matched as possible to the target reference. This phase often requires considerable time expenditure and expert supervision; moreover, if one considers that calibration is carried out both following the production of the diagnostic instrument but also after several months of its use in the operational context, it is easy to deduce that automating this process on the one hand would improve the diagnostic yield, and on the other hand would reduce downtime and recalibration. The project aims to improve and automate the calibration process by introducing machine learning techniques for image classification. The results of the project find application on all instruments used for imaging, whether they are based on MRI, computed tomography, X-ray or ultrasound techniques. In fact, the goal is to relate the configuration parameters of the instrument to the images it produces in order to eliminate noise and artefacts produced by misconfiguration. Despite this, in the project we will consider as a case study the images produced by an ultrasound-based device produced by Echolight S.p.A. Medical devices produced by Echolight S.p.A. exploit images derived from ultrasound scans (B-Mode) to automatically identify anatomical reference targets (lumbar vertebrae bone interfaces of the L1-L4 tract and proximal femur bone interface). Once the regions of interest (ROIs) are identified, a proprietary algorithm evaluates the spectral characteristics of selected portions of the raw ultrasound signal related to the analyzed bone tissues. From the analysis of the raw signal characteristics, a measure of the bone mineral density (BMD) of the analyzed anatomical sites is determined. In order to provide reliable, repeatable, and accurate BMD measurements, special calibration and testing procedures have been developed, however, which require several manual measurements and checks, resulting in a high human-time commitment and, consequently, introducing a risk of human error on the collection and interpretation of the collected measurements and results. Leveraging the image processing and image classification techniques developed within the project, the algorithm will provide output indicative of the presence of artefacts or other alterations in the performance of the ultrasound system in production in order to possibly intervene with further modifications and calibrations. As part of the project, standard conditions for conducting tests will also be defined through the use of specific ultrasound phantoms provided by the company.

Prof. Italo Epicoco ([email protected]) is the scientific responsible for this research grant.

DEADLINE: June 24, 2022
ALL INCLUSIVE GROSS AMOUNT (for 18 months): 29050,50 euro (i.e., 19367 euro annual gross amount)

NOTE: Foreign candidates are strongly encouraged to contact Prof. Epicoco by email if they need help/support in order to prepare their application: he will be glad to assist.

Prof. Massimo Cafaro, Ph.D.
Associate Professor of Parallel Algorithms and Data Mining/Machine Learning
Head of HPC Lab https://hpc-lab.unisalento.it
Director of Master in Applied Data Science

Department of Engineering for Innovation
University of Salento, Lecce, Italy
Via per Monteroni
73100 Lecce, Italy

Voice/Fax +39 0832 297371

Web https://www.massimocafaro.it
Web https://www.unisalento.it/people/massimo.cafaro

E-mail [email protected]
E-mail [email protected]
E-mail [email protected]

INGV
National Institute of Geophysics and Volcanology
Via di Vigna Murata 605
Roma

CMCC Foundation
Euro-Mediterranean Center on Climate Change
Via Augusto Imperatore, 16 - 73100 Lecce
[email protected]

unisalento.it