Non-speech human sounds, such as coughs and snores, and their patterns are associated with different respiratory diseases, including asthma, chronic obstructive pulmonary disease (COPD), as well as other health difficulties such as sleep disorders. Thereby, researchers and physicians have been using coughs and snores as symptoms while reporting and assessing respiratory diseases, their stages, and sleep quality. However, so far, the assessments frequently depend on different types of patient-reported surveys, which inherently suffer from various limitations, such as recall biases, human errors. Therefore, automated detection and reporting of coughs and snores can improve the disease assessment and monitoring. In this paper, we present an automated approach to detect coughs and snores from smartphone-microphones using generalized, semi-personalized and personalized modeling schemes. We analyze three separate datasets and different combinations of three types of nocturnal noises (i.e., sounds from air conditioners (AC), dog barks, and sirens) using the Mel-frequency cepstral coefficient (MFCC) features and different classification techniques. We find that a generalized model with the support vector machine (SVM) classifier can achieve an average accuracy of 0.86 ± 0.14, F1 score of 0.86± 0.13, and area under the receiver operating characteristic curve (AUC-ROC) of 0.94 ± 0.08. These performances can further be improved to an average accuracy of 0.96± 0.08, F1 score of 0.96± 0.08, and AUC-ROC of 0.98 ± 0.04 using the personalized random forest (RF) model. The results show the potential for smartphones to automatically report symptoms of respiratory diseases as well as sleep disorders. Furthermore, we find that our models perform consistently well while testing on separate datasets in the presence of multiple background-noises.
{"title":"Nocturnal Cough and Snore Detection Using Smartphones in Presence of Multiple Background-Noises","authors":"Sudip Vhaduri","doi":"10.1145/3378393.3402273","DOIUrl":"https://doi.org/10.1145/3378393.3402273","url":null,"abstract":"Non-speech human sounds, such as coughs and snores, and their patterns are associated with different respiratory diseases, including asthma, chronic obstructive pulmonary disease (COPD), as well as other health difficulties such as sleep disorders. Thereby, researchers and physicians have been using coughs and snores as symptoms while reporting and assessing respiratory diseases, their stages, and sleep quality. However, so far, the assessments frequently depend on different types of patient-reported surveys, which inherently suffer from various limitations, such as recall biases, human errors. Therefore, automated detection and reporting of coughs and snores can improve the disease assessment and monitoring. In this paper, we present an automated approach to detect coughs and snores from smartphone-microphones using generalized, semi-personalized and personalized modeling schemes. We analyze three separate datasets and different combinations of three types of nocturnal noises (i.e., sounds from air conditioners (AC), dog barks, and sirens) using the Mel-frequency cepstral coefficient (MFCC) features and different classification techniques. We find that a generalized model with the support vector machine (SVM) classifier can achieve an average accuracy of 0.86 ± 0.14, F1 score of 0.86± 0.13, and area under the receiver operating characteristic curve (AUC-ROC) of 0.94 ± 0.08. These performances can further be improved to an average accuracy of 0.96± 0.08, F1 score of 0.96± 0.08, and AUC-ROC of 0.98 ± 0.04 using the personalized random forest (RF) model. The results show the potential for smartphones to automatically report symptoms of respiratory diseases as well as sleep disorders. Furthermore, we find that our models perform consistently well while testing on separate datasets in the presence of multiple background-noises.","PeriodicalId":176951,"journal":{"name":"Proceedings of the 3rd ACM SIGCAS Conference on Computing and Sustainable Societies","volume":"26 3","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132546230","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
As mobile internet growth continues to bring New Internet Users (NIUs) online, technology has adapted to fit this user segment. User barriers like devices and connectivity have declined as mobile phone prices have become more affordable and infrastructure has continued to develop, connecting more communities globally. App development has also evolved to better suit users on low-cost Android devices. Lite apps have entered the space as a solution for users in constrained environments. While there are many benefits to lite app designs, their effectiveness is unclear for their likely target beneficiaries: NIUs coming online. In this mixed-method study we explore the experience for NIUs trying out a smartphone with lite apps for a month in Brazil and India (n=62). We conducted this research by collecting diary data and follow-up in-person interviews. Results found that three phases of challenges occurred in the first 28 days with a lite smartphone: 1) getting started with accounts, 2) learning how to use the mobile platform and apps, and 3) meeting expectations and mastering the internet. Through understanding the friction points in each phase, insights surfaced design principles for future NIU technology.
{"title":"28 Days Later: New Internet Users in Brazil and India Try a Lite Smartphone for a Month","authors":"Jennifer Zamora","doi":"10.1145/3378393.3402247","DOIUrl":"https://doi.org/10.1145/3378393.3402247","url":null,"abstract":"As mobile internet growth continues to bring New Internet Users (NIUs) online, technology has adapted to fit this user segment. User barriers like devices and connectivity have declined as mobile phone prices have become more affordable and infrastructure has continued to develop, connecting more communities globally. App development has also evolved to better suit users on low-cost Android devices. Lite apps have entered the space as a solution for users in constrained environments. While there are many benefits to lite app designs, their effectiveness is unclear for their likely target beneficiaries: NIUs coming online. In this mixed-method study we explore the experience for NIUs trying out a smartphone with lite apps for a month in Brazil and India (n=62). We conducted this research by collecting diary data and follow-up in-person interviews. Results found that three phases of challenges occurred in the first 28 days with a lite smartphone: 1) getting started with accounts, 2) learning how to use the mobile platform and apps, and 3) meeting expectations and mastering the internet. Through understanding the friction points in each phase, insights surfaced design principles for future NIU technology.","PeriodicalId":176951,"journal":{"name":"Proceedings of the 3rd ACM SIGCAS Conference on Computing and Sustainable Societies","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134444470","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Santiago Correa, Noman Bashir, Andrew Tran, David E. Irwin, Jay Taneja
The means of electrifying households and the resulting electricity networks are rapidly evolving. Traditionally, an extension of existing centralized grids was the only prominent technique, but now electrification is seeing massive expansion via decentralized solar home systems (SHSs). These systems consist of a low-wattage photovoltaic (PV) panel (typically 5-100W), a battery, a collection of energy-efficient DC appliances, and a charge controller. Spurred by significant advances and reduced costs in solar, batteries, energyefficient appliances, and mobile money-driven business models, SHSs have proliferated rapidly, with tens of millions of systems now deployed, primarily in regions with otherwise low rates of electricity access. In this work, we profile a large deployment of solar home systems in Western Kenya to ascertain the dominant generation and consumption patterns.We note that there are often substantial mismatches between generation and consumption, and that PV overgeneration presents an opportunity via networking of households. We consider the opportunity to leverage system interconnection to enable increased connectivity among households, challenging typical electricity system architecture by effectively creating ad hoc electricity grids at the edges of the overall electricity network. Further, we consider the potential to integrate households without SHSs ("passive nodes") into these electricity networks, as a low-cost opportunity to increase electrification rates. Considering energy curtailment, the spatial distribution of households, and infrastructure costs, we build a decision problem for interconnecting existing SHSs with passive nodes. Our analysis shows that compared to the all-SHS solutions that are presently achieving widespread deployment, we show that interconnecting existing SHSs can increase electrification rates by more than 25% and reduce average costs by up to 30% per household.
{"title":"Extend: A Framework for Increasing Energy Access by Interconnecting Solar Home Systems","authors":"Santiago Correa, Noman Bashir, Andrew Tran, David E. Irwin, Jay Taneja","doi":"10.1145/3378393.3402288","DOIUrl":"https://doi.org/10.1145/3378393.3402288","url":null,"abstract":"The means of electrifying households and the resulting electricity networks are rapidly evolving. Traditionally, an extension of existing centralized grids was the only prominent technique, but now electrification is seeing massive expansion via decentralized solar home systems (SHSs). These systems consist of a low-wattage photovoltaic (PV) panel (typically 5-100W), a battery, a collection of energy-efficient DC appliances, and a charge controller. Spurred by significant advances and reduced costs in solar, batteries, energyefficient appliances, and mobile money-driven business models, SHSs have proliferated rapidly, with tens of millions of systems now deployed, primarily in regions with otherwise low rates of electricity access. In this work, we profile a large deployment of solar home systems in Western Kenya to ascertain the dominant generation and consumption patterns.We note that there are often substantial mismatches between generation and consumption, and that PV overgeneration presents an opportunity via networking of households. We consider the opportunity to leverage system interconnection to enable increased connectivity among households, challenging typical electricity system architecture by effectively creating ad hoc electricity grids at the edges of the overall electricity network. Further, we consider the potential to integrate households without SHSs (\"passive nodes\") into these electricity networks, as a low-cost opportunity to increase electrification rates. Considering energy curtailment, the spatial distribution of households, and infrastructure costs, we build a decision problem for interconnecting existing SHSs with passive nodes. Our analysis shows that compared to the all-SHS solutions that are presently achieving widespread deployment, we show that interconnecting existing SHSs can increase electrification rates by more than 25% and reduce average costs by up to 30% per household.","PeriodicalId":176951,"journal":{"name":"Proceedings of the 3rd ACM SIGCAS Conference on Computing and Sustainable Societies","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131365284","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Corporate Social Responsibility (CSR) is seen as a means for companies to contribute towards broader societal goals beyond their immediate industrial focus, and companies are known to donate a part of their profits to social development for education, health, and other sectors. In 2014, the Government of India made CSR mandatory for companies beyond a certain level of profitability. It was observed however that many geographies in need of financial assistance for social development actually did not receive much CSR funds. In this paper, we investigate what might be the reasons behind how companies choose the locations for their CSR investments. In particular, we examine political reasons where companies may use CSR to seek favours from politicians. We find several interesting patterns and show that there might be grounds for the government to regulate CSR to some extent.
{"title":"What Drives Location Preference for Corporate Social Responsibility (CSR) Investments in India?","authors":"Varun Pareek, Rohit Sharma, Anirban Sen, Arundeep Gupta, Manikaran Kathuria, Aaditeshwar Seth","doi":"10.1145/3378393.3402263","DOIUrl":"https://doi.org/10.1145/3378393.3402263","url":null,"abstract":"Corporate Social Responsibility (CSR) is seen as a means for companies to contribute towards broader societal goals beyond their immediate industrial focus, and companies are known to donate a part of their profits to social development for education, health, and other sectors. In 2014, the Government of India made CSR mandatory for companies beyond a certain level of profitability. It was observed however that many geographies in need of financial assistance for social development actually did not receive much CSR funds. In this paper, we investigate what might be the reasons behind how companies choose the locations for their CSR investments. In particular, we examine political reasons where companies may use CSR to seek favours from politicians. We find several interesting patterns and show that there might be grounds for the government to regulate CSR to some extent.","PeriodicalId":176951,"journal":{"name":"Proceedings of the 3rd ACM SIGCAS Conference on Computing and Sustainable Societies","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124620621","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In the 2018-19 influenza season, between 37.4 and 42.9 million people in the United States experienced flu like symptoms. From that number, 431 to 647 thousand were hospitalized and 36400 to 61200 (most of them children and seniors) succumbed to the disease. Due to the annual mutation of the very many strands of the flu virus, new vaccines must be developed and administered every flu season. Therefore, the prediction of the rate of growth in reported infection cases of each strand of the flu is paramount to ensuring the correct supply of vaccines per strand. A great tool for making future predictions using existing data is Machine learning - specifically Neural Networks. eVision (Epidemic Vision) is a software using Long Short-Term Memory (LSTM) neural networks under research and development by Santa Clara University's EPIC (Ethical, Pragmatic, and Intelligent Computing) and Bioinnovation & Design labs to predict the trend of influenza cases throughout the flu season using data from the CDC, WHO, and Google Trends in order to help pharmaceuticals decide on the ramping up or down of their development of tester kits, vaccines, and medicines weeks in advance.
{"title":"Influenza Forecasting","authors":"Navid Shaghaghi, Andrés Calle, George Kouretas","doi":"10.1145/3378393.3402286","DOIUrl":"https://doi.org/10.1145/3378393.3402286","url":null,"abstract":"In the 2018-19 influenza season, between 37.4 and 42.9 million people in the United States experienced flu like symptoms. From that number, 431 to 647 thousand were hospitalized and 36400 to 61200 (most of them children and seniors) succumbed to the disease. Due to the annual mutation of the very many strands of the flu virus, new vaccines must be developed and administered every flu season. Therefore, the prediction of the rate of growth in reported infection cases of each strand of the flu is paramount to ensuring the correct supply of vaccines per strand. A great tool for making future predictions using existing data is Machine learning - specifically Neural Networks. eVision (Epidemic Vision) is a software using Long Short-Term Memory (LSTM) neural networks under research and development by Santa Clara University's EPIC (Ethical, Pragmatic, and Intelligent Computing) and Bioinnovation & Design labs to predict the trend of influenza cases throughout the flu season using data from the CDC, WHO, and Google Trends in order to help pharmaceuticals decide on the ramping up or down of their development of tester kits, vaccines, and medicines weeks in advance.","PeriodicalId":176951,"journal":{"name":"Proceedings of the 3rd ACM SIGCAS Conference on Computing and Sustainable Societies","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123039607","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Rose Nakasi, Ernest Mwebaze, A. Zawedde, J. Tusubira, Gilbert Maiga
While microscopy diagnosis through supervised learning for image analysis notably contributes to malaria detection, it has limitations. Among its principle challenges is the manual and tiresome process of data annotation for the classification task. The manual annotation of data is prone to inaccuracy defects due to bias, subjectivity and unclear images resulting into many false positives. This is normally due to personal independent judgements that vary from individual microscopists hence summatively affecting the accuracy of the model. In this paper, we seek to investigate the possibility of classifying the negative far examples and the positive near examples from the positives in thick blood smear images for malaria detection. Assessing the classification performance could potentially inform us of the quality of training dataset and guide n selecting the best training dataset for a malaria parasite detection task. We employ the Mean Squared Error (MSE) to distinguish between positive and negative images. We later investigate the performance of the VGG-16 classification model based on how close or far negative examples are from positives. Experimental results showed that negative examples far from the positives produce better results than those near and that the proposed method could potentially be used to reduce false positives and bias in the training data.
{"title":"An Approach for Assessing Quality of Labeled Data for a Machine Learning Task in Malaria Detection","authors":"Rose Nakasi, Ernest Mwebaze, A. Zawedde, J. Tusubira, Gilbert Maiga","doi":"10.1145/3378393.3402265","DOIUrl":"https://doi.org/10.1145/3378393.3402265","url":null,"abstract":"While microscopy diagnosis through supervised learning for image analysis notably contributes to malaria detection, it has limitations. Among its principle challenges is the manual and tiresome process of data annotation for the classification task. The manual annotation of data is prone to inaccuracy defects due to bias, subjectivity and unclear images resulting into many false positives. This is normally due to personal independent judgements that vary from individual microscopists hence summatively affecting the accuracy of the model. In this paper, we seek to investigate the possibility of classifying the negative far examples and the positive near examples from the positives in thick blood smear images for malaria detection. Assessing the classification performance could potentially inform us of the quality of training dataset and guide n selecting the best training dataset for a malaria parasite detection task. We employ the Mean Squared Error (MSE) to distinguish between positive and negative images. We later investigate the performance of the VGG-16 classification model based on how close or far negative examples are from positives. Experimental results showed that negative examples far from the positives produce better results than those near and that the proposed method could potentially be used to reduce false positives and bias in the training data.","PeriodicalId":176951,"journal":{"name":"Proceedings of the 3rd ACM SIGCAS Conference on Computing and Sustainable Societies","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124432017","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The City of Cape Town (CoCT) experienced three years of drought, which necessitated the implementation of water demand management strategies by one of the universities in CoCT to reduce water consumption. This study used persuasive system (persuasive information campaign (PIC) disseminated to students using Short Messaging Service (SMS), email and both SMS and email) in three residences and tested its effectiveness in increasing students' intention to save water. The extended Theory of Planned Behaviour and Partial Least Square Path Modelling were used for data collection and analysis. The factor loading showed that students who received the PIC by both SMS and email were the most persuaded to increase their intention to save water. Overall, PIC significantly influenced students' attitude towards water-saving, and students' attitude was the strongest predictor of intention to save water. This study highlights the importance of persuasive system in encouraging the sustainable use of scarce natural resources.
{"title":"Persuasive information campaign to save water in Universities: An option for water-stressed areas?","authors":"J. Azaki, U. Rivett","doi":"10.1145/3378393.3402238","DOIUrl":"https://doi.org/10.1145/3378393.3402238","url":null,"abstract":"The City of Cape Town (CoCT) experienced three years of drought, which necessitated the implementation of water demand management strategies by one of the universities in CoCT to reduce water consumption. This study used persuasive system (persuasive information campaign (PIC) disseminated to students using Short Messaging Service (SMS), email and both SMS and email) in three residences and tested its effectiveness in increasing students' intention to save water. The extended Theory of Planned Behaviour and Partial Least Square Path Modelling were used for data collection and analysis. The factor loading showed that students who received the PIC by both SMS and email were the most persuaded to increase their intention to save water. Overall, PIC significantly influenced students' attitude towards water-saving, and students' attitude was the strongest predictor of intention to save water. This study highlights the importance of persuasive system in encouraging the sustainable use of scarce natural resources.","PeriodicalId":176951,"journal":{"name":"Proceedings of the 3rd ACM SIGCAS Conference on Computing and Sustainable Societies","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116700105","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Emily L. Aiken, Guadalupe Bedoya, Aidan Coville, J. Blumenstock
Recent papers demonstrate that non-traditional data, from mobile phones and other digital sensors, can be used to roughly estimate the wealth of individual subscribers. This paper asks a question more directly relevant to development policy: Can non-traditional data be used to more efficiently target development aid? By combining rich survey data from a "big push" anti-poverty program in Afghanistan with detailed mobile phone logs from program beneficiaries, we study the extent to which machine learning methods can accurately differentiate ultra-poor households eligible for program benefits from other households deemed ineligible. We show that supervised learning methods leveraging mobile phone data can identify ultra-poor households as accurately as standard survey-based measures of poverty, including consumption and wealth; and that combining survey-based measures with mobile phone data produces classifications more accurate than those based on a single data source. We discuss the implications and limitations of these methods for targeting extreme poverty in marginalized populations.
{"title":"Targeting Development Aid with Machine Learning and Mobile Phone Data: Evidence from an Anti-Poverty Intervention in Afghanistan","authors":"Emily L. Aiken, Guadalupe Bedoya, Aidan Coville, J. Blumenstock","doi":"10.1145/3378393.3402274","DOIUrl":"https://doi.org/10.1145/3378393.3402274","url":null,"abstract":"Recent papers demonstrate that non-traditional data, from mobile phones and other digital sensors, can be used to roughly estimate the wealth of individual subscribers. This paper asks a question more directly relevant to development policy: Can non-traditional data be used to more efficiently target development aid? By combining rich survey data from a \"big push\" anti-poverty program in Afghanistan with detailed mobile phone logs from program beneficiaries, we study the extent to which machine learning methods can accurately differentiate ultra-poor households eligible for program benefits from other households deemed ineligible. We show that supervised learning methods leveraging mobile phone data can identify ultra-poor households as accurately as standard survey-based measures of poverty, including consumption and wealth; and that combining survey-based measures with mobile phone data produces classifications more accurate than those based on a single data source. We discuss the implications and limitations of these methods for targeting extreme poverty in marginalized populations.","PeriodicalId":176951,"journal":{"name":"Proceedings of the 3rd ACM SIGCAS Conference on Computing and Sustainable Societies","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128448930","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
To extract information at scale, researchers are increasingly applying semantic segmentation techniques to remotely-sensed imagery. While fully-supervised learning enables accurate pixelwise segmentation, compiling the exhaustive datasets required is often prohibitively expensive, and open-source datasets that do exists are frequently inexact and non-exhaustive. In this paper, we present a novel and generalizable two-stage framework that enables improved pixelwise image segmentation given misaligned and missing annotations. First, we introduce the Alignment Correction Network to rectify incorrectly registered open source labels. Next, we demonstrate a segmentation model - the Pointer Segmentation Network - that uses corrected labels to predict infrastructure footprints despite missing annotations. We demonstrate the transferability of our method to lower quality data sources by applying the Alignment Correction Network to correct OpenStreetMaps building footprints, and we show the accuracy of the Pointer Segmentation Network in predicting cropland boundaries in California. Overall, our methodology is robust for multiple applications with varied amounts of training data present, thus offering a method to extract reliable information from noisy, partial data.
{"title":"Learning to segment from misaligned and partial labels","authors":"Simone Fobi, Terence Conlon, Jay Taneja, V. Modi","doi":"10.1145/3378393.3402254","DOIUrl":"https://doi.org/10.1145/3378393.3402254","url":null,"abstract":"To extract information at scale, researchers are increasingly applying semantic segmentation techniques to remotely-sensed imagery. While fully-supervised learning enables accurate pixelwise segmentation, compiling the exhaustive datasets required is often prohibitively expensive, and open-source datasets that do exists are frequently inexact and non-exhaustive. In this paper, we present a novel and generalizable two-stage framework that enables improved pixelwise image segmentation given misaligned and missing annotations. First, we introduce the Alignment Correction Network to rectify incorrectly registered open source labels. Next, we demonstrate a segmentation model - the Pointer Segmentation Network - that uses corrected labels to predict infrastructure footprints despite missing annotations. We demonstrate the transferability of our method to lower quality data sources by applying the Alignment Correction Network to correct OpenStreetMaps building footprints, and we show the accuracy of the Pointer Segmentation Network in predicting cropland boundaries in California. Overall, our methodology is robust for multiple applications with varied amounts of training data present, thus offering a method to extract reliable information from noisy, partial data.","PeriodicalId":176951,"journal":{"name":"Proceedings of the 3rd ACM SIGCAS Conference on Computing and Sustainable Societies","volume":"40 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124036409","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Karan Ahuja, A. Bose, Mohit Jain, K. Dey, Anil Joshi, K. Achary, Blessin Varkey, Chris Harrison, Mayank Goel
Autism Spectrum Disorder (ASD) is a universal and often lifelong neuro-developmental disorder. Individuals with ASD often present comorbidities such as epilepsy, depression, and anxiety. In the United States, in 2014, 1 out of 68 people was affected by autism, but worldwide, the number of affected people drops to 1 in 160. This disparity is primarily due to underdiagnosis and unreported cases in resource-constrained environments. Wiggins et al. 1 found that, in the US, children of color are under-identified with ASD. Missing a diagnosis is not without consequences; approximately 26% of adults with ASD are under-employed, and are under-enrolled in higher education. Unfortunately, ASD diagnosis is not straightforward and involves a subjective assessment of the patient's behavior. Because such assessments can be noisy and even non-existent in low-resource environments, many cases go unidentified. Many such cases remain undiagnosed even when the patient reaches adolescence or adulthood. There is a need for an objective, low-cost, and ubiquitous approach to diagnose ASD. Autism is often characterized by symptoms such as limited interpersonal and social communication skills, and difficulty in face recognition and emotion interpretation. When watching video media, these symptoms can manifest as reduced eye fixation, resulting in characteristic gaze behaviors. Thus, we developed an approach to screen patients with ASD using their gaze behavior while they watch videos on a laptop screen. We used a dedicated eye tracker to record the participant's gaze. With data from 60 participants (35 with ASD and 25 without ASD), our algorithm demonstrates 92.5% classification accuracy after the participants watched 15 seconds of the video. We also developed a proof-of-concept regression model that estimates the severity of the condition and achieves a mean absolute error of 2.03 on the Childhood Autism Rating Scale (CARS). One of the most common approaches to identify individuals with ASD involves studying family home videos and investigating an infant's gaze and interactions with their families. However, having an expert carefully inspect hours of home video is expensive and unscalable. Our approach is more accessible and ubiquitous as we can directly sense the gaze of the user while they watch videos. Such sensing can be directly deployed on billions of smartphones around the world that are equipped with a front-facing camera. In our current exploration, we use a dedicated eye-tracker but achieving similar performance using an unmodified s martphone c amera is not far-fetched. Our results demonstrate that passively tracking a user's gaze pattern while they watch videos on a screen can enable robust identification of individuals with ASD. Past work has used specially-created visual content to detect ASD, but getting large sets of the population to watch specific videos is hard. Thus, we focus on generic content and selected four prosaic video scenes as a proof of con
{"title":"Gaze-based Screening of Autistic Traits for Adolescents and Young Adults using Prosaic Videos","authors":"Karan Ahuja, A. Bose, Mohit Jain, K. Dey, Anil Joshi, K. Achary, Blessin Varkey, Chris Harrison, Mayank Goel","doi":"10.1145/3378393.3402242","DOIUrl":"https://doi.org/10.1145/3378393.3402242","url":null,"abstract":"Autism Spectrum Disorder (ASD) is a universal and often lifelong neuro-developmental disorder. Individuals with ASD often present comorbidities such as epilepsy, depression, and anxiety. In the United States, in 2014, 1 out of 68 people was affected by autism, but worldwide, the number of affected people drops to 1 in 160. This disparity is primarily due to underdiagnosis and unreported cases in resource-constrained environments. Wiggins et al. 1 found that, in the US, children of color are under-identified with ASD. Missing a diagnosis is not without consequences; approximately 26% of adults with ASD are under-employed, and are under-enrolled in higher education. Unfortunately, ASD diagnosis is not straightforward and involves a subjective assessment of the patient's behavior. Because such assessments can be noisy and even non-existent in low-resource environments, many cases go unidentified. Many such cases remain undiagnosed even when the patient reaches adolescence or adulthood. There is a need for an objective, low-cost, and ubiquitous approach to diagnose ASD. Autism is often characterized by symptoms such as limited interpersonal and social communication skills, and difficulty in face recognition and emotion interpretation. When watching video media, these symptoms can manifest as reduced eye fixation, resulting in characteristic gaze behaviors. Thus, we developed an approach to screen patients with ASD using their gaze behavior while they watch videos on a laptop screen. We used a dedicated eye tracker to record the participant's gaze. With data from 60 participants (35 with ASD and 25 without ASD), our algorithm demonstrates 92.5% classification accuracy after the participants watched 15 seconds of the video. We also developed a proof-of-concept regression model that estimates the severity of the condition and achieves a mean absolute error of 2.03 on the Childhood Autism Rating Scale (CARS). One of the most common approaches to identify individuals with ASD involves studying family home videos and investigating an infant's gaze and interactions with their families. However, having an expert carefully inspect hours of home video is expensive and unscalable. Our approach is more accessible and ubiquitous as we can directly sense the gaze of the user while they watch videos. Such sensing can be directly deployed on billions of smartphones around the world that are equipped with a front-facing camera. In our current exploration, we use a dedicated eye-tracker but achieving similar performance using an unmodified s martphone c amera is not far-fetched. Our results demonstrate that passively tracking a user's gaze pattern while they watch videos on a screen can enable robust identification of individuals with ASD. Past work has used specially-created visual content to detect ASD, but getting large sets of the population to watch specific videos is hard. Thus, we focus on generic content and selected four prosaic video scenes as a proof of con","PeriodicalId":176951,"journal":{"name":"Proceedings of the 3rd ACM SIGCAS Conference on Computing and Sustainable Societies","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115844970","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：