Search

Your search keyword '"Barbarisi, Santiago"' showing total 3 results
Start Over Author "Barbarisi, Santiago" Search Limiters Full Text
3 results on '"Barbarisi, Santiago"'

2. Self-Knowledge Distillation for Surgical Phase Recognition

Author

Zhang, Jinglu, Barbarisi, Santiago, Kadkhodamohammadi, Abdolrahim, Stoyanov, Danail, and Luengo, Imanol

Subjects
Electrical Engineering and Systems Science - Image and Video Processing, Computer Science - Computer Vision and Pattern Recognition
Abstract

Purpose: Advances in surgical phase recognition are generally led by training deeper networks. Rather than going further with a more complex solution, we believe that current models can be exploited better. We propose a self-knowledge distillation framework that can be integrated into current state-of-the-art (SOTA) models without requiring any extra complexity to the models or annotations. Methods: Knowledge distillation is a framework for network regularization where knowledge is distilled from a teacher network to a student network. In self-knowledge distillation, the student model becomes the teacher such that the network learns from itself. Most phase recognition models follow an encoder-decoder framework. Our framework utilizes self-knowledge distillation in both stages. The teacher model guides the training process of the student model to extract enhanced feature representations from the encoder and build a more robust temporal decoder to tackle the over-segmentation problem. Results: We validate our proposed framework on the public dataset Cholec80. Our framework is embedded on top of four popular SOTA approaches and consistently improves their performance. Specifically, our best GRU model boosts performance by +3.33% accuracy and +3.95% F1-score over the same baseline model. Conclusion: We embed a self-knowledge distillation framework for the first time in the surgical phase recognition training pipeline. Experimental results demonstrate that our simple yet powerful framework can improve performance of existing phase recognition models. Moreover, our extensive experiments show that even with 75% of the training set we still achieve performance on par with the same baseline model trained on the full set.

Published
2023

3. Análisis del rendimiento académico de los estudiantes de Ingeniería de la UPF utilizando Data Science

Author

Barbarisi, Santiago

Subjects
Big data, Universitats -- Espanya -- Aspectes sociològics, Students performance, Educational data, Data science
Abstract

Treball de fi de grau en Sistemes Audiovisuals Tutor: Xavier Binefa Valls Hoy en día la Ciencia de Datos nos ofrece nuevos caminos para adquirir un mejor entendimiento de este mundo, nos ayuda a tomar decisiones sobre el futuro razonando sobre el pasado, y es, como indica la revista Harvard Business Review, ‘el trabajo más sexy del siglo XXI'. En este Trabajo de Fin de Grado exploramos los patrones de rendimiento presentes en los datos de los estudiantes de Ingeniería de la UPF a partir de dos perspectivas diferentes: por un lado analizamos los estudiantes y las características de acceso que los conforman, y por otro, estudiamos como los resultados de las asignaturas del primer trimestre se correlacionan con las grandes tasas de abandonos Universitarios. Este proyecto resulta significativo para comprender como las características de acceso de los estudiantes, como la vía de acceso, el centro de procedencia o el grado que realizan, afectan a los resultados que obtienen y a su trayectoria académica en la Universidad. Al mismo tiempo veremos como y por qué han sucedido los abandonos durante los últimos años, para establecer como la UPF ha reaccionado y gestionado estos casos, y de esta forma ofrecer guías para disminuir estos sucesos. Now days Data Science offers new ways to achieve a better comprehension of this world, it helps us to take better decisions about our future reasoning about our pass and, as Harvard Business Review suggests, it is ‘the sexiest job of the 21st Century’. In this Bachelor’s Degree Final Project we explore the patterns on the achievements of the Engineering’s students from the UPF, from two different perspectives: one analysing the students and the features of the enrolment, and another where we study how correlate are the subjects from the first period and the high rates of University drop out. The project is significant to understand how the enrolment features of students, such as the path of access, the center of origin or the degree they perform, affect the results they obtain and their academic career at the University. At the same time, we will see how and why the dropout occurred during the last years, to establish how the UPF has reacted and managed these cases, and thus offer guidelines to reduce these events.

Published
2017

Catalog

Books, media, physical & digital resources
See catalog results