@Article{BhalajiNagarajan2023,
author="Bhalaji Nagarajan
and Marc Bola{\~n}os
and Eduardo Aguilar
and Petia Radeva",
title="Deep ensemble-based hard sample mining for food recognition",
journal="Journal of Visual Communication and Image Representation",
year="2023",
volume="95",
pages="103905",
abstract="Deep neural networks represent a compelling technique to tackle complex real-world problems, but are over-parameterized and often suffer from over- or under-confident estimates. Deep ensembles have shown better parameter estimations and often provide reliable uncertainty estimates that contribute to the robustness of the results. In this work, we propose a new metric to identify samples that are hard to classify. Our metric is defined as coincidence score for deep ensembles which measures the agreement of its individual models. The main hypothesis we rely on is that deep learning algorithms learn the low-loss samples better compared to large-loss samples. In order to compensate for this, we use controlled over-sampling on the identified {\textquoteright}{\textquoteright}hard{\textquoteright}{\textquoteright} samples using proper data augmentation schemes to enable the models to learn those samples better. We validate the proposed metric using two public food datasets on different backbone architectures and show the improvements compared to the conventional deep neural network training using different performance metrics.",
optnote="MILAB",
optnote="exported from refbase (http://158.109.8.37/show.php?record=3844), last updated on Fri, 10 Nov 2023 11:17:39 +0100",
opturl="https://www.sciencedirect.com/science/article/pii/S1047320323001554"
}