Using Deep Learning for Collecting Data about Museum Visitor Behavior

Nowadays, technology makes it possible to admire objects and artworks exhibited all over the world remotely. We have been able to appreciate this convenience even more in the last period, in which the pandemic has forced us into our homes for a long time. However, visiting art sites in person remains a truly unique experience. Even during on-site visits, technology can help make them much more satisfactory, by assisting visitors during the fruition of cultural and artistic resources. To this aim, it is necessary to monitor the active user for acquiring information about their behavior. We, therefore, need systems able to monitor and analyze visitor behavior. The literature proposes several techniques for the timing and tracking of museum visitors. In this article, we propose a novel approach to indoor tracking that can represent a promising and non-expensive solution for some of the critical issues that remain. In particular, the system we propose relies on low-cost equipment (i.e., simple badges and off-the-shelf RGB cameras) and harnesses one of the most recent deep neural networks (i.e., Faster R-CNN) for detecting specific objects in an image or a video sequence with high accuracy. An experimental evaluation performed in a real scenario, namely, the “Exhibition of Fake Art” at Roma Tre University, allowed us to test our system on site. The collected data has proven to be accurate and helpful for gathering insightful information on visitor behavior.

Technological and Digital Interventions for Mental Health and Wellbeing: An Overview of Systematic Reviews

Background: Research increasingly shows how selective and targeted use of technology within care and welfare can have several advantages including improved quality of care and active user involvement.Purpose: The current overview of reviews aims to summarize the research on the effectiveness of technology for mental health and wellbeing. The goal is to highlight and structure the diverse combinations of technologies and interventions used so far, rather than to summarize the effectiveness of singular approaches.Methods: The current overview includes reviews published in the past five years with a focus on effectiveness of digital and technological interventions targeting mental health and wellbeing.Results: A total of 246 reviews could be included. All reviews examined the effectiveness of digital and technological interventions in the context of care and welfare. A combination of two taxonomies was created through qualitative analysis, based on the retrieved interventions and technologies in the reviews. Review classification shows a predominance of reviews on psychotherapeutic interventions using computers and smartphones. It is furthermore shown that when smartphone applications as stand-alone technology are researched, the primary focus is on self-help, and that extended reality is the most researched emerging technology to date.Conclusion: This overview of reviews shows that a wide range of interventions and technologies, with varying focus and target populations, have been studied in the field of care and wellbeing. The current overview of reviews is a first step to add structure to this rapidly changing field and may guide both researchers and clinicians in further exploring the evidence-base of particular approaches.

Трамп, Твиттер и фейковые новости (Trump, Twitter and Fake News)

This study aims to provide rationality regarding Donald Trump as a fake news aggregator. Donald Trump's leadership from 2017-2020 is controversial and created a massive wave of fake news. As a populist leader, he often issued statements that confused the public during his reign, causing people's trust in the Trump administration to decline. He made the statement not only on national television but also on social media. Social media is the right political communication funnel for any populist leader when it comes to audience reach. Donald Trump is an active user especially on Twitter and uses it to spread misinformation and disinformation to spread what he calls as a truth. Many statements make Donald Trump worthy of being called a fake news aggregator. This study uses a qualitative approach with the content analysis method. Thirty-two samples of Donald Trump's hoax statements that were examined found that the hoaxes spread by him were not limited to just one topic. This research has both theoretical and practical significance. From a theoretical point of view, this research contributes to the development of literature regarding the relationship between hoaxes and populist leaders. In practical terms, this literature contributes to understanding the characteristics of populists and how social media is used as a funnel to spread hoaxes.Keywords: Hoax; Donald Trump; Populism; social media; Twitter.

Syndromic Surveillance-Based Estimates of Vaccine Efficacy Against COVID-Like Illness from Emerging Omicron and COVID-19 Variants

New infections from the omicron variant of SARS-CoV-2 have been increasing dramatically in South Africa since first identification in November 2021. Despite increasing uptake of COVID-19 vaccine, there are concerns vaccine protection against omicron may be reduced compared to other variants. We sought to characterize a surrogate measure of vaccine efficacy in Gauteng, South Africa by leveraging real-time syndromic surveillance data. The University of Maryland Global COVID Trends and Impact Survey (UMD-CTIS) is an online, cross-sectional survey conducted among users sampled from the Facebook active user base. We derived three COVID-like illness (CLI) definitions (stringent, classic, and broad) using combinations of self-reported symptoms (present or not in the prior 24 hours) that broadly tracked with reported COVID-19 cases during June 18, 2021 - December 14, 2021 (inclusive of the delta wave and up-trend of the omicron wave). We used syndromic-surveillance-based CLI prevalence measures among the vaccinated (PV) and unvaccinated (PU) respondents to estimate VECLIP = 1 - (PV/PU), a proxy for vaccine efficacy, during the delta (June 18-July 18, N= 9,387 surveys) and omicron (December 4-14, N= 2,389 surveys) wave periods. We assume no waning immunity, CLI prevalence approximates incident infection with each variant, and vaccinated and unvaccinated survey respondents in the two variant wave periods are exchangeable. The vaccine appears to have consistently lower VECLIP against omicron, compared to delta, regardless of the CLI definition used. Stringent CLI (i.e. anosmia plus fever, cough and/or myalgias) yielded a delta VECLIP = 0.85 [0.54, 0.95] higher than omicron VECLIP = 0.62 [0.46, 0.72]. Classic CLI (cough plus anosmia, fever, and/or myalgias) gave lower estimates (delta VECLIP = 0.76 [0.54, 0.87], omicron VECLIP = 0.51 [0.42, 0.59]), but omicron was still lower than delta. We acknowledge the potential for measurement, confounding, and selection bias, as well as limitations for generalizability for these self-reported, syndromic surveillance-based VECLIP measures. Thus VECLIP as estimates of true, population-level vaccine efficacy should therefore be taken with caution. Nevertheless, these preliminary findings demonstrating declining VECLIP raise concern for a true decline in vaccine efficacy versus waning immunity as a potential contributor to the omicron variant taking hold in Gauteng and elsewhere.

Some New Perspectives on the Multi-user Detection in Uplink Grant-Free NOMA using Deep Neural Network

To support the explosive growth of the Internet of Things (IoT), Uplink (UL) grant-free Non-Orthogonal Multiple Access (NOMA) emerges as a promising technology. It has the potential of offering scalable and low-cost solutions for the resource-constrained Massive Machine Type Communication (mMTC) systems. In principle, the grant-free NOMA enables small signaling overhead and low access latency time by circumventing complicated grant-access based procedures which is commonly found in the legacy wireless networks. In a UL grant-free system, a complete Multi-User Detection (MUD) algorithm not only performs the Active User Detection (AUD) but also the Channel Estimation (CE) and the Data Detection (DD). By exploiting the naturally occurring sparse user activity in the mMTC systems, the MUD problem can be solved using a wide range of Compressive Sensing based algorithms (CS-MUD). However, some alternative routes have been explored in the literature as well. The utility of these algorithms, in general, revolve around some assumptions about the channel or the availability of perfect channel information at the Base Station (BS). How these assumptions are met in a practical circumstance is, however, an important concern. In this work we devise an end-to-end MUD using Deep Neural Network (DNN) where we relax these assumptions. We approximate an ensemble of trained DNN based MUD using Knowledge Distillation (KD) to enable fast AUD at the Base Station (BS). Furthermore, using the inter-resource correlation, we estimate the channels of the active users which is an ill-posed problem otherwise. We carry out elaborate numerical investigation to validate the efficacy of the proposed approach for the UL grant-free NOMA systems.

Hierarchical Multi-User Detection for Highly Overloaded Uplink Grant-Free NOMA

Grant-Free Non Orthogonal Multiple Access (NOMA) offers promising solutions to realize uplink (UL) massive Machine Type Communication (mMTC) using limited spectrum resources, while reducing signalling overhead. Because of the sparse, sporadic activities exhibited by the user equipments (UE), the active user detection (AUD) problem is often formulated as a compressive sensing problem. In line of that, greedy sparse recovery algorithms are spearheading the development of compressed sensing based multi-user detectors (CS-MUD). However, for a given number of resources, the performance of CS-MUD algorithms are fundamentally limited at higher overloading of NOMA. To circumvent this issue, in this work, we propose a two-stage hierarchical multi-user detection framework, where the UEs are randomly assigned to some pre-defined clusters. The active UEs split their data transmission frame into two phases. In the first phase an UE uses the sinusoidal spreading sequence (SS) of its cluster. In the second phase the UE uses its own unique random SS. At phase 1 of detection, the active clusters are detected, and a reduced sensing matrix is constructed. This matrix is used in Phase 2 to recover the active UE indices using some sparse recovery algorithm. Numerical investigations validate the efficacy of the proposed algorithm in highly overloaded scenarios.

Hierarchical Multi-User Detection for Highly Overloaded Uplink Grant-Free NOMA

Grant-Free Non Orthogonal Multiple Access (NOMA) offers promising solutions to realize uplink (UL) massive Machine Type Communication (mMTC) using limited spectrum resources, while reducing signalling overhead. Because of the sparse, sporadic activities exhibited by the user equipments (UE), the active user detection (AUD) problem is often formulated as a compressive sensing problem. In line of that, greedy sparse recovery algorithms are spearheading the development of compressed sensing based multi-user detectors (CS-MUD). However, for a given number of resources, the performance of CS-MUD algorithms are fundamentally limited at higher overloading of NOMA. To circumvent this issue, in this work, we propose a two-stage hierarchical multi-user detection framework, where the UEs are randomly assigned to some pre-defined clusters. The active UEs split their data transmission frame into two phases. In the first phase an UE uses the sinusoidal spreading sequence (SS) of its cluster. In the second phase the UE uses its own unique random SS. At phase 1 of detection, the active clusters are detected, and a reduced sensing matrix is constructed. This matrix is used in Phase 2 to recover the active UE indices using some sparse recovery algorithm. Numerical investigations validate the efficacy of the proposed algorithm in highly overloaded scenarios.

Some New Perspectives on the Multi-user Detection in Uplink Grant-Free NOMA using Deep Neural Network

To support the explosive growth of the Internet of Things (IoT), Uplink (UL) grant-free Non-Orthogonal Multiple Access (NOMA) emerges as a promising technology. It has the potential of offering scalable and low-cost solutions for the resource-constrained Massive Machine Type Communication (mMTC) systems. In principle, the grant-free NOMA enables small signaling overhead and low access latency time by circumventing complicated grant-access based procedures which is commonly found in the legacy wireless networks. In a UL grant-free system, a complete Multi-User Detection (MUD) algorithm not only performs the Active User Detection (AUD) but also the Channel Estimation (CE) and the Data Detection (DD). By exploiting the naturally occurring sparse user activity in the mMTC systems, the MUD problem can be solved using a wide range of Compressive Sensing based algorithms (CS-MUD). However, some alternative routes have been explored in the literature as well. The utility of these algorithms, in general, revolve around some assumptions about the channel or the availability of perfect channel information at the Base Station (BS). How these assumptions are met in a practical circumstance is, however, an important concern. In this work we devise an end-to-end MUD using Deep Neural Network (DNN) where we relax these assumptions. We approximate an ensemble of trained DNN based MUD using Knowledge Distillation (KD) to enable fast AUD at the Base Station (BS). Furthermore, using the inter-resource correlation, we estimate the channels of the active users which is an ill-posed problem otherwise. We carry out elaborate numerical investigation to validate the efficacy of the proposed approach for the UL grant-free NOMA systems.

Some New Perspectives on the Multi-user Detection in Uplink Grant-Free NOMA using Deep Neural Network

To support the explosive growth of the Internet of Things (IoT), Uplink (UL) grant-free Non-Orthogonal Multiple Access (NOMA) emerges as a promising technology. It has the potential of offering scalable and low-cost solutions for the resource-constrained Massive Machine Type Communication (mMTC) systems. In principle, the grant-free NOMA enables small signaling overhead and low access latency time by circumventing complicated grant-access based procedures which is commonly found in the legacy wireless networks. In a UL grant-free system, a complete Multi-User Detection (MUD) algorithm not only performs the Active User Detection (AUD) but also the Channel Estimation (CE) and the Data Detection (DD). By exploiting the naturally occurring sparse user activity in the mMTC systems, the MUD problem can be solved using a wide range of Compressive Sensing based algorithms (CS-MUD). However, some alternative routes have been explored in the literature as well. The utility of these algorithms, in general, revolve around some assumptions about the channel or the availability of perfect channel information at the Base Station (BS). How these assumptions are met in a practical circumstance is, however, an important concern. In this work we devise an end-to-end MUD using Deep Neural Network (DNN) where we relax these assumptions. We approximate an ensemble of trained DNN based MUD using Knowledge Distillation (KD) to enable fast AUD at the Base Station (BS). Furthermore, using the inter-resource correlation, we estimate the channels of the active users which is an ill-posed problem otherwise. We carry out elaborate numerical investigation to validate the efficacy of the proposed approach for the UL grant-free NOMA systems.