Research Assistant - Benguerir

PROFILE 1:

Topic: Adding interpretability into automatic crop classification using deep learning techniques

Description: Nowadays, considering the continuous population growth and the limited availability of food, it is necessary to monitor agricultural activities on a regular basis, so as to allow for increased efficiency in food production, while protecting natural ecosystems. In this context, crop classification can be used to provide information on production and thus become a useful tool for developing sustainable plans and reducing environmental problems associated with agriculture. Therefore, timely collection and analysis of data from large crop areas is of great interest. Traditionally, such analysis is carried out using computational tools and satellite image processing with artificial intelligence (AI) techniques-especially those based on deep learning. Although several efficient deep learning techniques (such as convolutional and recurrent neural networks) have emerged in the field of multispectral image analysis, the problem of crop classification still needs more accurate, and in-biological-context interpretable solutions.

This research assistantship is aimed at supporting PhD students on exploring and developing AI technologies that incorporate elements of molecular biology and OMICs in the study of crops in such a way that a good trade-off balance between accuracy and interpretability in crop classification tasks can be achieved.

Profile: 

• Strong biology background.
• Postgraduate training in precision agricultural settings. 
• Experience with software packages and tools for statistics and data visualization.
• (Initial) background on machine learning.

Duration: 6 Months.

PROFILE 2:

Topic: Design of kernelized formulations for interpretable neural-network-based data analysis approaches

Description: Kernel functions are highly versatile and powerful to analyze data. Broadly, kernels can both provide a graph-based representation through pairwise similarities and incorporate prior knowledge via functional analysis tools such as generalized inner products. Recent studies have proved that neural-network-driven approaches can accurately be represented by kernel machines.

The research assistant is expected to research on functional analysis and matrix algebra to pose kernelized formulations to represent modern machine learning (specially those based on neural networks) in such a manner that sharply defined concepts of both mathematical and in-domain/business-related interpretability can be incorporated. 

Profile: 

• Graduate degree as mathematician or computer scientist with a strong background on functional analysis, theoretical machine learning, and linear algebra.
• Master’s degree in theoretical computer science or applied mathematics.

Duration: 6 Months.

PROFILE 3:

Topic: Improving the resilience of olive cultivation to marginal pedoclimatic changes using simulation and data analysis

Description: Recently, the olive industry has undergone significant changes in farming practices, with a shift from traditional low-density to new high-density crop systems. Irrigation and the use of modern, efficient farming techniques have had a significant impact on the industry, with integrated production becoming increasingly important.

The research assistant is expected to research on simulation and data analysis on optimization of the fertilization-irrigation regime interaction for improving the resilience of olive cultivation to marginal pedoclimatic changes. 

Profile: 

• Graduate degree in environmental bio-analysis or related areas.
• Master’s degree in Biotechnology and AgroBioSciences.
• Experience in biological, biochemical, physico-chemical bioanalysis techniques, and software for simulation and data analysis.

Duration: 6 Months

PROFILE 4:

Topic: Using deep learning for metaverse-enabled educational applications

Description: Many users and experts at the field define metaverse as the way of extending the experience of watching at a screen by incorporating the capacity to navigate within 3D and to alter the points-of-view -enabling more interactive and realistic interactions. It became popular for video games but now is gaining an increasing interest in several settings: industry, medicine, and education. Besides, the rapid development of deep learning is enabling more and more interactive experiences to be built in the 

metaverse.

For this research assistantship position, the focus will be on educational applications by turning virtual reality scenarios into an educational environment powered by modern deep learning and computational technologies.

Duration: 12 Months.

Profile: 

• Graduate degree in IT.
• Strong background in deep learning (transfer learning, transformers and CNN), and computer vision (image filtering, transformation, segmentation and classification).
• Experience or willingness to learn at a fast pace virtual reality engines.
• (Desirable) Knowledge on technology-enabled educational paradigms.