Archivio della ricerca della Scuola Superiore Sant'Anna
Not a member yet
26957 research outputs found
Sort by
Toward Greener Matrix Operations by Lossless Compressed Formats
Get PDFSparse matrix-vector multiplication (SpMV) is a fundamental operation in machine learning, scientific computing, and graph algorithms. In this paper, we investigate the space, time, and energy efficiency of SpMV using various compressed formats for large sparse matrices, focusing specifically on Boolean matrices and real-valued vectors. Through extensive analysis and experimentation on server and edge devices, we observed that different matrix compression formats entail distinct trade-offs among space efficiency, execution time, and energy consumption. Notably, selecting the appropriate compressed format can reduce energy consumption by an order of magnitude on both servers and single-board computers. Moreover, our experiments reveal that while data parallelism can improve execution speed and energy efficiency, optimizing both simultaneously poses interesting challenges. Specifically, we demonstrate that for certain compression schemes, the optimal degree of parallelism for minimizing execution time does not coincide with that for energy efficiency. This finding aligns with recent results in the literature, thus challenging the prevailing assumption of a straightforward linear correlation between execution time and energy usage. Overall, our findings advocate for software engineers to adopt compressed data structures as important instruments in the development of more energy-efficient software components, extending beyond SpMV
Micropatterned Styrene–Butadiene–Styrene Thin Films Doped with Barium Titanate Nanoparticles: Effects on Myoblast Differentiation
Get PDFBiohybrid actuators exploit the contraction of biological components (muscle cells) to produce a force. In particular, bottom-up approaches use tissue engineering techniques, by coupling cells with a proper scaffold to obtain constructs undergoing contraction and guaranteeing actuation in biohybrid devices. However, the fabrication of actuators able to recapitulate the organization and maturity of native muscle is not trivial. In this field, quasi-two-dimensional (2D) substrates are raising interest due to their high surface/thickness ratio and the possibility of functionalizing their surface. In this work, we fabricated micropatterned thin films made of poly(styrene–butadiene–styrene) (SBS) doped with barium titanate nanoparticles (BTNPs) for fostering myogenic differentiation. We investigated material concentrations and fabrication process parameters to obtain thin microgrooved films with an average thickness below 1 μm, thus featured by a relatively low flexural rigidity and with an anisotropic topography to guide cell alignment and myotube formation. The embodiment of BTNPs did not significantly affect the film’s mechanical properties. Interestingly, the presence of BTNPs enhanced the expression of myogenic differentiation markers (i.e., MYH1, MYH4, MYH8, and ACTA1). The results show the promising potential of SBS thin films doped with BTNPs, opening avenues in the fields of biohybrid actuation and skeletal muscle tissue engineering
How should an explanation be? A mapping of technical and legal desiderata of explanations for machine learning models
Get PDFMachine learning (ML) systems are abundant in our world.
However, most of these systems are not understandable, which poses several challenges, including their safety, proper
functioning and accountability. Further, ML models are susceptible to social biases, which can lead to unjust and
discriminatory situations. The field of eXplainable Artificial Intelligence (XAI) attempts to answer these challenges by
providing explanation methods for ML models. However, there is still an open debate about the necessary desiderata of such methods, including the often-missing consideration of the legal side of explanation desiderata. In this work, we put forward a set of five technical and five legal desiderata of XAI and develop a multi-layered mapping encompassing the dynamics among and between the two sets and linking them to actual requirements. From the standpoint of legality, we rely on the European requirements explainability and justificability. In our mapping, we draw the interdependencies and the intersections between the technical and legal desiderata, creating an image that visualises
the assessment of the technical and legal driving forces (desiderata matching requirements) in the design and provision
of explanations. Ultimately, explainability and justificability desiderata must be systematic; understood as a dynamic, circular
and iterative process
Dispositivo autoiniettore
No full textIl dispositivo autoiniettore, in particolare per la somministrazione di iniezioni sottocutanee o intramuscolari di una sostanza, comprende un involucro (2); un gruppo di azionamento (4) comprendente un organo di attivazione (13) associato
a detto involucro (2), detto organo di attivazione (13) 5 essendo atto ad essere sottoposto, in uso, a una forza esercitata da un utilizzatore lungo una direzione longitudinale. Il dispositivo comprende altresì un contenitore (3), disposto all’interno di detto involucro (2), che è associato in corrispondenza di una estremità ad almeno un ago (6), detto contenitore (3) conformando al suo interno una 10 camera (5) atta a contenere una sostanza
AGAINST ADMINISTRATIVE LAW. RAGIONI, ARGOMENTI E IMPLICAZIONI DI UN DIBATTITO STATUNITENSE
No full textThe Last Decade in Cardiac Amyloidosis Advances in Understanding Pathophysiology, Diagnosis and Quantification, Prognosis, Treatment Strategies, and Monitoring Response
Get PDF
The dynamics of automation adoption: Firm-level heterogeneity and aggregate employment effects
No full textWe investigate the impact of investment in automation-related goods on adopting and non-adopting firms in the Italian economy during 2011–2019. We integrate datasets on trade activities, firms’, and workers’ characteristics for the population of Italian importing firms and estimate the effects on adopters’ outcomes within a difference-in-differences design exploiting import lumpiness in product categories linked to automation technologies (including robots). We find a positive average adoption effect on the adopters’ employment: firms are, on average, around 3% larger in terms of employment after an automation spike. Crucially, the employment effect is heterogeneous across firms: a positive effect is predominant among small firms, which are around 5% larger five years after the spike; on the contrary, a negative displacement effect is predominant among medium and large firms, with an employment contraction at five years of around −4%. This result can shed light on one potential reason behind the mixed results found in previous studies, which often rely on samples with different size distributions. We complete the framework with a 5-digit sector-level analysis showing that adopting automation technologies has an overall weak negative effect on aggregate employment, and with an analysis of the competition effects of automation, showing that non-adopters suffer a loss in sales and employment
