Zero-Shot Learning and its Applications from Autonomous Vehicles to COVID-19 Diagnosis: A Review

27 Pages Posted: 11 Jun 2020 Last revised: 22 Jun 2020

See all articles by Mahdi Rezaei

Mahdi Rezaei

University of Leeds, Institute for Transport Studies; The University of Auckland, School of Computer Science

Mahsa Shahidi

Independent; Qazvin Azad University

Date Written: June 10, 2020

Abstract

The challenge of learning a new concept, object, or a new medical disease recognition without receiving any examples beforehand is called Zero-Shot Learning (ZSL). One of the major issues in deep learning-based methodologies such as in Medical Imaging, Autonomous Systems and other real-world applications is the requirement of feeding a large annotated and labelled datasets, prepared by an expert human to train the network model. ZSL is known for having minimal human intervention by mainly relying only on previously known concepts and current auxiliary information. This is ever-growing research for the cases where we have very limited or no datasets available and at the same time, the detection/recognition system has human-like characteristics in learning new concepts. Therefore, it makes it applicable in real-world scenarios, from developing autonomous vehicles to medical imaging and COVID-19 Chest X-Ray (CXR) based diagnosis.

In this review paper, we present the definition of the problem, we review over fundamentals, and the challenging steps of Zero-Shot Learning, including state-of-the-art categories of solutions as well as our recommended solution, motivations behind each approach, and their advantages over each category to guide the researchers to proceed with the best techniques and practices based on their applications. Inspired from different settings and extensions, we introduce a novel and broaden solution called one$\slash$few-shot learning. We then review through different image datasets inducing medical and non-medical images, the variety of splits, and the evaluation protocols proposed so far. Finally, we discuss the recent applications and future directions of ZSL. We aim to convey a useful intuition through this paper towards the goal of handling complex computer vision learning tasks more similar to the way humans learn. We mainly focus on two applications in the current modern yet challenging era: copping with an early and fast diagnosis of COVID-19 cases in the 2020 pandemic, as well as helping to develop safer autonomous systems, such as self-driving cars and ambulances.

Keywords: Zero-shot learning; Chest X-ray (CXR); Machine Learning; Semantic Embedding; Deep Learning; COVID-19 Pandemic; SARS-CoV-2; Autonomous Vehicles; Supervised Annotation

Suggested Citation

Rezaei, Mahdi and Shahidi, Mahsa, Zero-Shot Learning and its Applications from Autonomous Vehicles to COVID-19 Diagnosis: A Review (June 10, 2020). Available at SSRN: https://ssrn.com/abstract=3624379 or http://dx.doi.org/10.2139/ssrn.3624379

Mahdi Rezaei (Contact Author)

University of Leeds, Institute for Transport Studies ( email )

34-40 University Raod, Institute for Transport Stu
Leeds
United Kingdom
01133435342 (Phone)

HOME PAGE: http://https://environment.leeds.ac.uk/transport/staff/9408/dr-mahdi-rezaei

The University of Auckland, School of Computer Science ( email )

Private Bag 92019
Auckland Mail Centre
Auckland, 1142
New Zealand

HOME PAGE: http://MahdiRezaei.auckland.ac.nz

Mahsa Shahidi

Independent ( email )

Qazvin Azad University ( email )

Nokhbegan Boulevard, Barajin
Qazvin, 10
Iran

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
25
Abstract Views
1,113
PlumX Metrics