Cláudia Brito

Dr. Cláudia Brito

orcid

Research Focus & Metrics

Academic Career Length

10Years (2017-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications

27 March 2023
Verified
Conference Paper
Open access
CBCláudia Brito
Cláudia Brito
Pedro G. FerreiraPedro G. Ferreira
Pedro G. Ferreira
BPBernardo Portela
Bernardo Portela
RORui Oliveira
Rui Oliveira
JPJoão Paulo
João Paulo
We propose Soteria, a system for distributed privacy-preserving Machine Learning (ML) that leverages Trusted Execution Environments (e.g. Intel SGX) to run code in isolated containers (enclaves). Unlike previous work, where all ML-related computation is performed at trusted enclaves, we introduce a hybrid scheme, combining computation done inside and outside these enclaves. The conducted experimental evaluation validates that our approach reduces the runtime of ML algorithms by up to 41%, when compared to previous related work. Our protocol is accompanied by a security proof, as well as a discussion regarding resilience against a wide spectrum of ML attacks.

1 January 2019
Verified
Book Chapter
Open access
PDF file
CBCláudia Brito
Cláudia Brito
MAMachado Ana
Machado Ana
SASousa António
Sousa António
When dealing with electrocardiography (ECG) the main focus relies on the classification of the heart's electric activity and deep learning has been proving its value over the years classifying the heartbeats, exhibiting great performance when doing so. Following these assumptions, we propose a deep learning model based on a ResNet architecture with convolutional 1D layers to classify the beats into one of the 4 classes: normal, atrial premature contraction, premature ventricular contraction and others. Experimental results with MIT-BIH Arrhythmia Database confirmed that the model is able to perform well, obtaining an accuracy of 96% when using stochastic gradient descent (SGD) and 83% when using adaptive moment estimation (Adam), SGD also obtained F1-scores over 90% for the four classes proposed. A larger dataset was created and tested as unforeseen data for the trained model, proving that new tests should be done to improve the accuracy of it.

27 December 2024 in Transportation Engineering
Verified
Journal Article
Open access
CBCláudia Brito
Cláudia Brito
NPNoela Pina
Noela Pina
TETânia Esteves
Tânia Esteves
RVRicardo Vitorino
Ricardo Vitorino
ICInês Cunha
Inês Cunha
JPJoão Paulo
João Paulo

LBLaura da Conceição Mendes Brito
Laura da Conceição Mendes Brito
BCBeatriz Cepa
Beatriz Cepa
CBCláudia Brito
Cláudia Brito
ÂLÂngela Leite
Ângela Leite
MPM. Graça Pereira
M. Graça Pereira
Alzheimer’s disease (AD) places a profound global challenge, driven by its escalating prevalence and the multifaceted strain it places on individuals, families, and societies. Family caregivers (FCs), who are pivotal in supporting family members with AD, frequently endure substantial emotional, physical, and psychological demands. To better understand the determinants of family caregiving strain, this study employed machine learning (ML) to develop predictive models identifying factors that contribute to caregiver burden over time. Participants were evaluated across sociodemographic clinical, psychophysiological, and psychological domains at baseline (T1; N = 130), six months (T2; N = 114), and twelve months (T3; N = 92). Results revealed three distinct risk profiles, with the first focusing on T2 data, highlighting the importance of distress, forgiveness, age, and heart rate variability. The second profile integrated T1 and T2 data, emphasizing additional factors like family stress. The third profile combined T1 and T2 data with sociodemographic and clinical features, underscoring the importance of both assessment moments on distress at T2 and forgiveness at T1 and T2, as well as family stress at T1. By employing computational methods, this research uncovers nuanced patterns in caregiver burden that conventional statistical approaches might overlook. Key drivers include psychological factors (distress, forgiveness), physiological markers (heart rate variability), contextual stressors (familial dynamics, sociodemographic disparities). The insights revealed enable early identification of FCs at higher risk of burden, paving the way for personalized interventions. Such strategies are urgently needed as AD rates rise globally, underscoring the imperative to safeguard both patients and the caregivers who support them.

13 November 2023 in IEEE Access
Verified
Journal Article
Open access
CBCláudia Brito
Cláudia Brito
Pedro G. FerreiraPedro G. Ferreira
Pedro G. Ferreira
BPBernardo L. Portela
Bernardo L. Portela
RORui C. Oliveira
Rui C. Oliveira
JPJoão T. Paulo
João T. Paulo
The adoption of third-party machine learning (ML) cloud services is highly dependent on the security guarantees and the performance penalty they incur on workloads for model training and inference. This paper explores security/performance trade-offs for the distributed Apache Spark framework and its ML library. Concretely, we build upon a key insight: in specific deployment settings, one can reveal carefully chosen non-sensitive operations (e.g. statistical calculations). This allows us to considerably improve the performance of privacy-preserving solutions without exposing the protocol to pervasive ML attacks. In more detail, we propose Soteria, a system for distributed privacy-preserving ML that leverages Trusted Execution Environments (e.g. Intel SGX) to run computations over sensitive information in isolated containers (enclaves). Unlike previous work, where all ML-related computation is performed at trusted enclaves, we introduce a hybrid scheme, combining computation done inside and outside these enclaves. The experimental evaluation validates that our approach reduces the runtime of ML algorithms by up to 41% when compared to previous related work. Our protocol is accompanied by a security proof and a discussion regarding resilience against a wide spectrum of ML attacks.