Non-tenure-track/postdoctoral Caserta, Italy Apply
Job
Università degli studi della Campania "Luigi Vanvitelli"
Mathematics and Physics
Viale Lincoln 5
81100 Caserta
Italy
The address could not be found.

Department of Mathematics and Physics of Unicampania opens 4 postdoctoral positions (Assegni di ricerca) in Numerical Analysis, Optimization, Computer Science, Geometry and Statistics. 

 

If you need more info and details about the application you may contact

 

  1. Valentina De Simone (valentina.desimone@unicampania.it)
  2. Francesco Giannino (francesco.giannino@unina.it)
  3. Gerardo Toraldo (gerardo.toraldo@unicampania.it)
  4. Ferdinando Zullo (ferdinando.zullo@unicampania.it)

 

POSITION 1 - Development of  models, algorithms and software for glyph and incision extraction

 

Tutor: Valentina De Simone

Period: 21 Months

Deadline: 18 January 2023

Starting date: 1 March 2024

Areas: Numerical Analysis, Computer Science, Geometry

Monthly salary (gross): 1800€

 

Description of the activities

The project aims to develop new automated tools for archaeologists for extracting glyphs and inscriptions from archaeological artifacts in Sardinia, typical of the prehistoric and protohistoric periods. Currently, this extraction process is done manually and can be time-consuming. The starting point will be the analysis and development of three-dimensional models of objects using photos taken in various lighting conditions and viewpoints, employing techniques such as stereo-photometry and stereo-vision. Based on these models, the project intends to provide efficient algorithms and software for the extraction, classification, and interpretation of data contained in the artifacts.

 

POSITION 2 - Data analysis for glyph and incision extraction

 

Tutor: Valentina De Simone

Period: 12 Months

Deadline: 18 January 2023

Starting date: 1 March 2024

Areas: Statistics, Numerical Analysis, Geometry

Monthly salary (gross): 1600€

 

Description of the activities

The project aims to develop new automated tools for archaeologists to analyze artifacts, with a focus on creating three-dimensional models from photographs taken under different lighting conditions and perspectives, using techniques like photometric stereo and stereo-vision. These models will serve as the foundation for the extraction and analysis of inscriptions and glyphs, which are currently done manually and known to be labor-intensive and time-consuming. To achieve this, the project intends to develop machine learning algorithms for artifact classification, automating these processes and enabling further analysis. The methodology involves the use of classical techniques such as Linear Discriminant Analysis and Principal Component Analysis, along with modern deep learning techniques for image classification, text and pattern analysis, and shape recognition.

 

POSITION 3 - Algorithms of image analysis for vegetation pattern recognition

 

Tutor: Gerardo Toraldo

Period: 21 Months

Deadline: 18 January 2023

Starting date: 1 March 2024

Areas: Numerical Analysis, Computer Science

Monthly salary (gross): 1700€ 

 

Description of the activities

The aim of the research is to define the methods for Vegetation pattern (VP) recognition from the available satellite images and data. Such analysis must be computed over space and time in order not only to detect the most probable vegetation pattern of the considered geographical area, but to understand the correlation between images and other meaningful data (e.g., precipitation data, etc.). Then images must be pre-processed to improve their quality and then analysed with Machine Learning based approaches to detect the most likelihood vegetation pattern associated with the analysed area. Both supervised and unsupervised approaches can be explored as well as different pre-processing techniques will be employed.  A further possible objective of the contract is also the construction of an evaluation testbench for pre-processing and recognition methods. Such a framework must be able to allow its user to test and compare different combination of “pipeline”, to automatically tune their parameters and to automatically compose elementary blocks and evaluating their performance on a specified dataset.  The final aim is to construct an open framework able to attract the expertise of other researchers to obtain a continuous improvement of the recognition accuracy.

 

POSITION 4 - Design of inexact deterministic optimization algorithms

 

Tutor: Gerardo Toraldo

Period: 12 Months

Deadline: 18 January 2023

Starting date: 1 March 2024

Areas: Numerical Analysis, Optimization, Computer Science

Monthly salary (gross): 1600€ 

 

Description of the activities

The project will deal with First-order methods for large-scale minimization problems, possibly with bound or linear constraints, mainly focusing on step-length update rules exploiting the use of variable metrics related, possibly, to a diagonal matrix. The basic idea is to scale each component of the gradient by better capturing the local geometry of the problem. Better algorithm performances can be obtained alternating in an adaptive way using different step-length techniques, like the Barzilai-Borwein rules and more general limited-memory strategies, on the basis of their behavior during the iterative process and the accuracy of the iterates. A critical issue the research will address is the use of approximate gradient and projection operators for bound and linearly constrained large-scale nonlinear optimization problems. This paves the way toward the application of deterministic gradient methods to larger and larger problems, while attempting to preserve the well-studied properties of these methods.


Application Instructions

Since the call is in Italian the easisiest thing to do is to contact one among

 

  1. Valentina De Simone (valentina.desimone@unicampania.it)
  2. Francesco Giannino (francesco.giannino@unina.it)
  3. Gerardo Toraldo (gerardo.toraldo@unicampania.it)
  4. Ferdinando Zullo (ferdinando.zullo@unicampania.it)

(Add to Calendar)
This employer is not accepting applications through MathHire.org. Please follow the instructions above and refer to MathHire.org in your application.
Contact Person
Prof. Ferdinando Zullo