Istituto Italiano di Tecnologia

About: Istituto Italiano di Tecnologia is a facility organization based out in Genoa, Italy. It is known for research contribution in the topics: Humanoid robot & Robot. The organization has 4561 authors who have published 14595 publications receiving 437558 citations. The organization is also known as: Italian Institute of Technology & IIT.

Coding observed motor acts: different organizational principles in the parietal and premotor cortex of humans.

Abstract: Understanding actions of conspecifics is a fundamental social ability depending largely on the activation of a parieto-frontal network. Using functional MRI (fMRI), we studied how goal-directed movements (i.e., motor acts) performed by others are coded within this network. In the first experiment, we presented volunteers with video clips showing four different motor acts (dragging, dropping, grasping, and pushing) performed with different effectors (foot, hand, and mouth). We found that the coding of observed motor acts differed between premotor and parietal cortex. In the premotor cortex, they clustered according to the effector used, and in the inferior parietal lobule (IPL), they clustered according to the type of the observed motor act, regardless of the effector. Two subsequent experiments in which we directly contrasted these four motor acts indicated that, in IPL, the observed motor acts are coded according to the relationship between agent and object: Movements bringing the object toward the agent (grasping and dragging) activate a site corresponding approximately to the ventral part of the putative human AIP (phAIP), whereas movements moving the object away from the agent (pushing and dropping) are clustered dorsally within this area. These data provide indications that the phAIP region plays a role in categorizing motor acts according to their behavioral significance. In addition, our results suggest that in the case of motor acts typically done with the hand, the representations of such acts in phAIP are used as templates for coding motor acts executed with other effectors.

Bilevel Programming for Hyperparameter Optimization and Meta-Learning

Abstract: We introduce a framework based on bilevel programming that unifies gradient-based hyperparameter optimization and meta-learning. We show that an approximate version of the bilevel problem can be solved by taking into explicit account the optimization dynamics for the inner objective. Depending on the specific setting, the outer variables take either the meaning of hyperparameters in a supervised learning problem or parameters of a meta-learner. We provide sufficient conditions under which solutions of the approximate problem converge to those of the exact problem. We instantiate our approach for meta-learning in the case of deep learning where representation layers are treated as hyperparameters shared across a set of training episodes. In experiments, we confirm our theoretical findings, present encouraging results for few-shot learning and contrast the bilevel approach against classical approaches for learning-to-learn.

Green-synthesized W- and Mo-doped BiVO4 oriented along the {040} facet with enhanced activity for the sun-driven water oxidation

Abstract: A green hydrothermal synthesis approach, which employs almost neutral pH conditions and ammonium carbonate as a structural directing agent, has been used for the preparation of W- and Mo-doped BiVO4 powders with a monoclinic scheelite-like structure and preferential growth along the {0 4 0} facet. The crystal structure and morphology of the samples have been studied by means of XRD and FESEM analyses. The proper introduction of the W and Mo dopants in the lattice of BiVO4 crystals has been confirmed from XRD and XPS analyses. Moreover, it has been shown that the superficial amount of W and Mo dopants (obtained by means of XPS) plays a more important role than the overall (bulk and surface) dopant quantities in the BiVO4 (estimated by means of ICPMS analysis). The best of the so prepared materials demonstrated an extraordinary enhancement of the water oxidation rate under simulated sunlight (i.e. 942 μmol/g.h O2), which is about two-fold and 100 times higher than those of a state-of-art Mo-doped and the non-doped BiVO4 reference materials, respectively. The here reported BiVO4 photocatalysts are highly promising for different applications such as renewable energy production systems (i.e. solar fuels production form water and sunlight) and to resolve environmental issues (i.e. photocatalytic degradation of organic pollutants, water remediation, among others).

Progress in the Light Emission of Colloidal Semiconductor Nanocrystals

Abstract: Some 25 years ago it was found that semiconductor nanocrystals emitted light. Since then tremendous progress has been made with respect to increasing the emission quantum yields, extending the spectral range that may be addressed, from the UV across to the near infrared, and improving the color purity. Here some major lines in these developments are reviewed, touching on milestones as well as on the principles of the most successful preparative approaches.