Search Results for author: Sergio Decherchi

Found 8 papers, 3 papers with code

Label Critic: Design Data Before Models

1 code implementation5 Nov 2024 Pedro R. A. S. Bassi, Qilong Wu, Wenxuan Li, Sergio Decherchi, Andrea Cavalli, Alan Yuille, Zongwei Zhou

Label Critic can also check the label quality of a single AI Label with 71. 8% accuracy when no alternatives are available for comparison, prompting radiologists to review and edit if the estimated quality is low (19% depending on body structures).

A Hybrid Federated Kernel Regularized Least Squares Algorithm

no code implementations24 Jul 2024 Celeste Damiani, Yulia Rodina, Sergio Decherchi

Federated learning is becoming an increasingly viable and accepted strategy for building machine learning models in critical privacy-preserving scenarios such as clinical settings.

Federated Learning Privacy Preserving

Explanation is All You Need in Distillation: Mitigating Bias and Shortcut Learning

no code implementations13 Jul 2024 Pedro R. A. S. Bassi, Andrea Cavalli, Sergio Decherchi

We found that it is possible to train a neural network with explanation (e. g by Layer Relevance Propagation, LRP) distillation only, and that the technique leads to high resistance to shortcut learning, surpassing group-invariant learning, explanation background minimization, and alternative distillation techniques.

Language Modelling

Faster ISNet for Background Bias Mitigation on Deep Neural Networks

1 code implementation16 Jan 2024 Pedro R. A. S. Bassi, Sergio Decherchi, Andrea Cavalli

Representing a potentially massive training speed improvement over ISNet, the proposed architectures introduce LRP optimization into a gamut of applications that the original model cannot feasibly handle.

Conceptron: a probabilistic deep one-class classification method

no code implementations29 Sep 2021 Erika Gardini, Andrea Cavalli, Sergio Decherchi

One-class learning through deep architectures is a particularly challenging task; in this scenario the crasis of kernel methods and deep networks can represent a viable strategy to empower already effective methods.

Classification One-Class Classification

Distributed Kernel K-Means for Large Scale Clustering

no code implementations9 Oct 2017 Marco Jacopo Ferrarotti, Sergio Decherchi, Walter Rocchia

However, the systematic application of the kernelized version of k-means is hampered by its inherent square scaling in memory with the number of samples.

Clustering Computational chemistry

Cannot find the paper you are looking for? You can Submit a new open access paper.