-
RobCaps: Evaluating the Robustness of Capsule Networks against Affine Transformations and Adversarial Attacks
Authors:
Alberto Marchisio,
Antonio De Marco,
Alessio Colucci,
Maurizio Martina,
Muhammad Shafique
Abstract:
Capsule Networks (CapsNets) are able to hierarchically preserve the pose relationships between multiple objects for image classification tasks. Other than achieving high accuracy, another relevant factor in deploying CapsNets in safety-critical applications is the robustness against input transformations and malicious adversarial attacks.
In this paper, we systematically analyze and evaluate dif…
▽ More
Capsule Networks (CapsNets) are able to hierarchically preserve the pose relationships between multiple objects for image classification tasks. Other than achieving high accuracy, another relevant factor in deploying CapsNets in safety-critical applications is the robustness against input transformations and malicious adversarial attacks.
In this paper, we systematically analyze and evaluate different factors affecting the robustness of CapsNets, compared to traditional Convolutional Neural Networks (CNNs). Towards a comprehensive comparison, we test two CapsNet models and two CNN models on the MNIST, GTSRB, and CIFAR10 datasets, as well as on the affine-transformed versions of such datasets. With a thorough analysis, we show which properties of these architectures better contribute to increasing the robustness and their limitations. Overall, CapsNets achieve better robustness against adversarial examples and affine transformations, compared to a traditional CNN with a similar number of parameters. Similar conclusions have been derived for deeper versions of CapsNets and CNNs. Moreover, our results unleash a key finding that the dynamic routing does not contribute much to improving the CapsNets' robustness. Indeed, the main generalization contribution is due to the hierarchical feature learning through capsules.
△ Less
Submitted 25 April, 2023; v1 submitted 8 April, 2023;
originally announced April 2023.
-
Digital Enablers Of Construction Project Governance
Authors:
Paolo Eugenio Demagistris,
Sandro Petruzzi,
Rodolfo Pampaloni,
Milan Šmigić,
Alberto De Marco,
Waseem Khan,
Filippo Maria Ottaviani
Abstract:
Construction project governance relies on agreements between the actors along the construction industry value chain. The mutual obligations arising from these contracts rely on timely monetary transactions. Despite the advantages of automation in payment systems and improved access to digital progress data, several payment applications rely nonetheless on inefficient and time-consuming procedures…
▽ More
Construction project governance relies on agreements between the actors along the construction industry value chain. The mutual obligations arising from these contracts rely on timely monetary transactions. Despite the advantages of automation in payment systems and improved access to digital progress data, several payment applications rely nonetheless on inefficient and time-consuming procedures and documentation. This study examines the present technological advancements that can lead to fix this problem. A smart contract-based approach is ideal for managing construction progress payments that support autonomous process, it fills the gap between payments and project site progress evaluations. This article offers a way for automating construction payments by formalizing smart contracts execution on a decentralized block-chain-based system.
△ Less
Submitted 12 May, 2022;
originally announced May 2022.
-
COBOL to Java and Newspapers Still Get Delivered
Authors:
Alessandro De Marco,
Valentin Iancu,
Ira Asinofsky
Abstract:
This paper is an experience report on migrating an American newspaper company's business-critical IBM mainframe application to Linux servers by automatically translating the application's source code from COBOL to Java and converting the mainframe data store from VSAM KSDS files to an Oracle relational database. The mainframe application had supported daily home delivery of the newspaper since 197…
▽ More
This paper is an experience report on migrating an American newspaper company's business-critical IBM mainframe application to Linux servers by automatically translating the application's source code from COBOL to Java and converting the mainframe data store from VSAM KSDS files to an Oracle relational database. The mainframe application had supported daily home delivery of the newspaper since 1979. It was in need of modernization in order to increase interoperability and enable future convergence with newer enterprise systems as well as to reduce operating costs. Testing the modernized application proved to be the most vexing area of work. This paper explains the process that was employed to test functional equivalence between the legacy and modernized applications, the main testing challenges, and lessons learned after having operated and maintained the modernized application in production over the last eight months. The goal of delivering a functionally equivalent system was achieved, but problems remained to be solved related to new feature development, business domain knowledge transfer, and recruiting new software engineers to work on the modernized application.
△ Less
Submitted 10 August, 2018;
originally announced August 2018.
