-
Replay-free Online Continual Learning with Self-Supervised MultiPatches
Authors:
Giacomo Cignoni,
Andrea Cossu,
Alex Gomez-Villa,
Joost van de Weijer,
Antonio Carta
Abstract:
Online Continual Learning (OCL) methods train a model on a non-stationary data stream where only a few examples are available at a time, often leveraging replay strategies. However, usage of replay is sometimes forbidden, especially in applications with strict privacy regulations. Therefore, we propose Continual MultiPatches (CMP), an effective plug-in for existing OCL self-supervised learning str…
▽ More
Online Continual Learning (OCL) methods train a model on a non-stationary data stream where only a few examples are available at a time, often leveraging replay strategies. However, usage of replay is sometimes forbidden, especially in applications with strict privacy regulations. Therefore, we propose Continual MultiPatches (CMP), an effective plug-in for existing OCL self-supervised learning strategies that avoids the use of replay samples. CMP generates multiple patches from a single example and projects them into a shared feature space, where patches coming from the same example are pushed together without collapsing into a single point. CMP surpasses replay and other SSL-based strategies on OCL streams, challenging the role of replay as a go-to solution for self-supervised OCL.
△ Less
Submitted 13 February, 2025;
originally announced February 2025.
-
PD-L1 Classification of Weakly-Labeled Whole Slide Images of Breast Cancer
Authors:
Giacomo Cignoni,
Cristian Scatena,
Chiara Frascarelli,
Nicola Fusco,
Antonio Giuseppe Naccarato,
Giuseppe Nicoló Fanelli,
Alina Sîrbu
Abstract:
Specific and effective breast cancer therapy relies on the accurate quantification of PD-L1 positivity in tumors, which appears in the form of brown stainings in high resolution whole slide images (WSIs). However, the retrieval and extensive labeling of PD-L1 stained WSIs is a time-consuming and challenging task for pathologists, resulting in low reproducibility, especially for borderline images.…
▽ More
Specific and effective breast cancer therapy relies on the accurate quantification of PD-L1 positivity in tumors, which appears in the form of brown stainings in high resolution whole slide images (WSIs). However, the retrieval and extensive labeling of PD-L1 stained WSIs is a time-consuming and challenging task for pathologists, resulting in low reproducibility, especially for borderline images. This study aims to develop and compare models able to classify PD-L1 positivity of breast cancer samples based on WSI analysis, relying only on WSI-level labels. The task consists of two phases: identifying regions of interest (ROI) and classifying tumors as PD-L1 positive or negative. For the latter, two model categories were developed, with different feature extraction methodologies. The first encodes images based on the colour distance from a base color. The second uses a convolutional autoencoder to obtain embeddings of WSI tiles, and aggregates them into a WSI-level embedding. For both model types, features are fed into downstream ML classifiers. Two datasets from different clinical centers were used in two different training configurations: (1) training on one dataset and testing on the other; (2) combining the datasets. We also tested the performance with or without human preprocessing to remove brown artefacts Colour distance based models achieve the best performances on testing configuration (1) with artefact removal, while autoencoder-based models are superior in the remaining cases, which are prone to greater data variability.
△ Less
Submitted 15 April, 2024;
originally announced April 2024.
-
Retracing and assessing the CEP project
Authors:
Giovanni A. Cignoni,
Fabio Gadducci
Abstract:
The last decade witnessed a renewed interest in the development of the Italian computer industry and in the role of the Fifties pioneers in Rome, Milan, Ivrea, and Pisa. The aim of the paper is to retrace some steps of the CEP project, carried out by the University of Pisa in collaboration with Olivetti, by reassessing the documents preserved in the University archives. The project was a seminal e…
▽ More
The last decade witnessed a renewed interest in the development of the Italian computer industry and in the role of the Fifties pioneers in Rome, Milan, Ivrea, and Pisa. The aim of the paper is to retrace some steps of the CEP project, carried out by the University of Pisa in collaboration with Olivetti, by reassessing the documents preserved in the University archives. The project was a seminal enterprise for Italy, and among its accomplishments it delivered in 1957 the first Italian computer. The mix of public sector funding and industrial foretelling witnessed by the project is one of the leading examples in Italy of best practices, and its success paved the way for the birth of Computer Science in the country as an industry as well as a scientific discipline.
△ Less
Submitted 25 March, 2019;
originally announced April 2019.
-
Discovering Eastern European PCs by hacking them. Today
Authors:
Stefano Bodrato,
Fabrizio Caruso,
Giovanni A. Cignoni
Abstract:
Computer science would not be the same without personal computers. In the West the so called PC revolution started in the late '70s and has its roots in hobbyists and do-it-yourself clubs. In the following years the diffusion of home and personal computers has made the discipline closer to many people. A bit later, to a lesser extent, yet in a similar way, the revolution took place also in East Eu…
▽ More
Computer science would not be the same without personal computers. In the West the so called PC revolution started in the late '70s and has its roots in hobbyists and do-it-yourself clubs. In the following years the diffusion of home and personal computers has made the discipline closer to many people. A bit later, to a lesser extent, yet in a similar way, the revolution took place also in East European countries. Today, the scenario of personal computing has completely changed, however the computers of the '80s are still objects of fascination for a number of retrocomputing fans who enjoy using, programming and hacking the old 8-bits. The paper highlights the continuity between yesterday's hobbyists and today's retrocomputing enthusiasts, particularly focusing on East European PCs. Besides the preservation of old hardware and software, the community is engaged in the development of emulators and cross compilers. Such tools can be used for historical investigation, for example to trace the origins of the BASIC interpreters loaded in the ROMs of East European PCs.
△ Less
Submitted 21 January, 2019;
originally announced January 2019.
-
A Model for Configuration Management of Open Software Systems
Authors:
Roberto Barbuti,
Giovanni A. Cignoni,
Paolo Milazzo
Abstract:
The article proposes a model for the configuration management of open systems. The model aims at validation of configurations against given specifications. An extension of decision graphs is proposed to express specifications. The proposed model can be used by software developers to validate their own configurations across different versions of the components, or to validate configurations that in…
▽ More
The article proposes a model for the configuration management of open systems. The model aims at validation of configurations against given specifications. An extension of decision graphs is proposed to express specifications. The proposed model can be used by software developers to validate their own configurations across different versions of the components, or to validate configurations that include components by third parties. The model can also be used by end-users to validate compatibility among different configurations of the same application. The proposed model is first discussed in some application scenarios and then formally defined. Moreover, a type discipline is given to formally define validation of a configuration against a system specification
△ Less
Submitted 24 November, 2010;
originally announced November 2010.
