Ruggiero Seccia

Ruggiero Seccia
Visiting Student
330 North Orchard Street
Madison WI 53715
Room 4241C-1
seccia@wisc.edu
Preferred pronouns: he/him/his
Optimization algorithms for Machine Learning

Years at WID

2019 - present

About

Ruggiero Seccia was born in Rome, Italy, and has organized summer camps for children since the age of 14.

Education

M.Sc., Management Engineering, University of Rome La Sapienza

Research Description

So far, I have been working on optimization methods for speeding up the training phase of Machine Learning models. In particular, I have been focused on the definition of a general framework for the application of Block Coordinate decomposition schemes in training Deep Feedforward Neural Networks. This framework, by exploiting the layered structure of the model, has revealed to be able to reduce the training time needed to reach a good generalization error of the model. Furthermore, I have also worked on the application of Sequential Model-Based Optimization algorithms for improving the curriculum learning phase in Reinforcement Learning applications. Right now, in collaboration with professor S. Wright, I am working on the definition of a new algorithm for solving Linear Programming Problems by solving the KKT system associated with the primal-dual solution.

Affiliations

Department of Computer, Control and Management Engineering "Antonio Ruberti", DIAG, University of Rome "La Sapienza"