On April 26, 2013, Tatiana Tommasi successfully defended her PhD thesis entitled "Learning to Learn by Exploiting Prior Knowledge"
The work described in her thesis takes place in the context of visual recognition and category detection, in all scenarios where there is a need for learning a new category model but it is not possible or expensive, to obtain large amounts of annotated training data. The main contribution of the thesis of Tatiana Tommasi is to cast the problem of learning to learn from small samples into the max-margin classifiers framework, providing to the community principled algorithms for open ended learning of object categories from few samples. The power and generality of the results achieved in the thesis has further been demonstrated in other domains, such as control of prosthetic hands for amputees.
For more details click here: Learning to Learn by Exploiting Prior Knowledge