Internship Predictive Network Modeling Research

Rough network data being typically unlabeled limits application of neural networks to unsupervised learning tasks (learning from unlabeled data). Tools for automated network data labeling becomes thus a first-level priority to enable supervised learning tasks (learning from hypothesis/candidate models) by means of neural networks. These neural networks are primarily those involved in nonlinear function approximation underlying automated pattern detection/extraction and recognition/identification such as multi-dimensional interaction modeling and their synthetic representation. In the context of communication networks, for instance, labeled datasets can be used for training neural models to augment -or even replace- current local routing and forwarding process.

On the other hand, a label can be considered as encoding the value of the objective function underlying the machine learning task or the parameter of a model (structured learning). Hence, automated labeling implies that either this objective or the model are formally known. This operation can be realized either in batch or online mode. The former typically assumes that the main performance criteria relates to the size of the input (assumed to be complete) that needs to be processed to reach a given accuracy level. The latter implies time-constrained processing of incoming input following different arrival processes; thus, the number of cycles/inter-arrival times required to reach a given accuracy level.

Objective: automatic labeling method of network data, including spatio-temporal data (traffic, etc.), structural data (topological, etc.)

Task: blends modeling, algorithmic and experimental research; the candidate will have the opportunity to explore, evaluate and compare different labeling algorithms by formal and numerical methods. These tasks will be realized under supervision of senior (postdoc-level) researcher.

Duration: from 6 months to 1 year (max.)

Candidate profile:

• if MSc thesis: the candidate must be following the last year of the curriculum in, e.g., Applied mathematics, Math engineering, Theoretical computer science, Computer science engineering. Detailed coordinates of MSc promotor and his/her academic affiliation must be provided in the CV application form.
• if internship: the candidate must have completed his MSc (in one of these disciplines). Copy of the MSc diploma/certificate shall be included in annex of the CV. The internship can also be considered as part of post-MSc graduation or PhD graduation program.
• good knowledge of traffic/network simulation is considered as a plus.

Note well: candidate must have obtained their University degree from an academic institution of one of the EU country.

Starting date: March1st, 2021