Talal Shaikh - Academia.edu (original) (raw)

Papers by Talal Shaikh

Expert Systems, Aug 8, 2023

Expert systems are being extensively used to make critical decisions involving emotional analysis... more Expert systems are being extensively used to make critical decisions involving emotional analysis in affective computing. The evolution of deep learning algorithms has improved the potential for extracting value from multimodal emotional data. However, these black‐box algorithms do not often explain the heuristics behind processing the input features for achieving certain outputs. This study focuses on the risks of using black‐box deep learning models for critical tasks, such as emotion recognition, and describes how human understandable interpretations of the workings of these models are extremely important. This study utilizes one of the largest multimodal datasets available–CMU‐MOSEI. Many researchers have used the pre‐extracted features provided by the CMU Multimodal SDK with black‐box deep learning models making it difficult to interpret the contribution of its individual features. This study describes the implications of significant features from various modalities (audio, video, text) identified using XAI in Multimodal Emotion Recognition. It describes the process of curating reduced feature models by using the Gradient SHAP XAI method. These reduced models with highly contributing features achieve comparable and at times even better results compared to their corresponding all‐feature models as well as the baseline model GraphMFN. This study reveals that carefully selecting significant features for a model can help filter out irrelevant features, and attenuate the noise or bias caused by them, leading to an improved performance efficiency of the expert systems by making them transparent, easily interpretable, and trustworthy.

Lecture Notes in Networks and Systems

Lecture Notes in Networks and Systems

2018 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops), 2018

The rise of mobile phone usage across the world has given rise to many context aware applications... more The rise of mobile phone usage across the world has given rise to many context aware applications. To study the effects of context-aware application we have created a prototype application for the main demographics in a University. The prototype provides location based context-awareness in two facets: outside the university campus (outdoor) and inside the university campus (indoor). We have used Wi-Fi fingerprinting technology to achieve indoor positioning. Furthermore, the prototype is connected to Moodle, an online e-learning service via web services to provide course materials to the student based on the user, location (classroom) and time. The students can further interact with their peers spatially for discussions, as their location data is securely shared with their group mates. Our research evaluates the effect of context-awareness for two user groups; the students and the lecturers. The usage of both indoor and outdoor positioning within a single prototype, to provide the location aspect of context in our research, shows 94% acceptance rate between both the user groups.

2018 2nd International Conference on Smart Grid and Smart Cities (ICSGSC), 2018

Growing population, increasing urbanization and competition for shrinking resources, Food Securit... more Growing population, increasing urbanization and competition for shrinking resources, Food Security is a challenge around the world. The UN has made it a top priority in their Global Goals to promote sustainable agricultural practices with modern technologies and fair distribution system to sustain the world’s population. In this paper we investigate the implementation of a micro controlled environment agriculture system, the MicroCEA, a smart device that automates growing vegetables in an urban indoor residential setting. We have constructed and tested the device and the platform in the United Arab Emirates where about 80% of the food is imported.Our urban smart farming device is based on an extension of the MIT Food Computer version 1, which was adapted to the local environment based on the materials and equipment that could be locally sourced here. To develop our prototype the action research methodology was employed to leverage findings and to improve the next stage of the prototype development. We present our research into the design and testing of a prototyped Internet of Things (IoT) MicroCEA device after comparing other similar open source systems. A new IoT architecture for urban smart farming devices, a set of lessons learned that can be used in forwarding this field of study that can contribute toward achieving food security and improved nutrition and promote sustainable agriculture.

The relative influence of Semantic Web on Artificial Intelligence and vice versa is ever increasi... more The relative influence of Semantic Web on Artificial Intelligence and vice versa is ever increasing. However Artificial Intelligence and Semantic Web applications and research are going in different tangents as mentioned by the pioneers of Semantic Web. The real benefit would be on mixing semantic data for use in Artificially Intelligent apps so that reasoning can be done on open datasets for different entities. One way to do reasoning between entities would be by finding commonalities. that purpose we have created EnSense (Enhanced Sensing) — A web application that can find commonalities between two RDF objects using Semantic Web. We have devised an algorithm that could perform the task with greater speed, better number of relationships and quality as opposed to its closely related works. Our results have proven an 85% decrease in time taken and have 145% increase in the number of commonalities found. We provide a visual representation of the commonalities based on an interactive g...

Human emotion is one of the fundamental ways we express our feelings and currently the main focus... more Human emotion is one of the fundamental ways we express our feelings and currently the main focus is only facial expression for emotion extraction from an image. The aim of this paper is to enlighten the importance of other different features of an image in extracting the emotions apart from only facial features. The created system is able to accept an image uploaded by the user and detect the emotion individually. Firstly, the emotion is predicted based on the facial feature, then by the color and then by the text. Lastly, all three detected emotion is put to a comparison and a final output is displayed to the user. The created system went through an evaluation and the results were striking to see that the system prediction was often similar to the human prediction and the users were satisfied with the outcome.

2019 International Conference on Computational Intelligence and Knowledge Economy (ICCIKE), 2019

In recent years, data-driven (image based) methodologies like deep learning and computer vision h... more In recent years, data-driven (image based) methodologies like deep learning and computer vision have made computers immensely accurate in terms of identifying features inside images. Research in this area has given way to a relatively new set of deep learning models known as Generative models which generate images alongside identifying features inside them. These models, particularly conditional generative adversarial networks (CGANs), conditional variational autoencoders (CVAE) and generative stochastic networks (GSN) have become popular as they are able to translate images from one setting to another while keeping the structure of generated images aligned with the input images. In this paper, we review the work that has been done using these models to the area of web design automation which needs to be considered during the development phase. We also try to identify the benefits of implementing these models based on architectural features while keeping in view the different proble...

2017 International Conference on Computer and Applications (ICCA), 2017

Medical data contain very valuable information which can save many lives if it is analyzed and ut... more Medical data contain very valuable information which can save many lives if it is analyzed and utilized efficiently. Efficient analysis of this large volume of data demands the right choice of predictors and this in turn can impact the accuracy of the decision support system. Dimensionality reduction and feature subset selection are two techniques to reduce the number of features used in classification. In this paper we perform an empirical evaluation of four feature selection methods when applied in conjunction with Random Forest classifier. The feature selection techniques applied are Relief feature selection algorithm, Random forest selector, Recursive feature elimination and Boruta Feature selection algorithm. Results show that feature selection methods boosts the performance of the classifiers and in this case the features selected by the Boruta feature selection algorithm gives the best results.

Proceedings of the International Conference on Big Data and Advanced Wireless Technologies - BDAW '16, 2016

For a typical smart city many IoT (Internet of Things) architectures have been proposed to integr... more For a typical smart city many IoT (Internet of Things) architectures have been proposed to integrate the various ICT Solutions. Since there are many perspectives, the emphasis and the design of each architecture vary. So at the moment there is no standard architecture. The best approach is to adapt architecture for its intended use. Most of the current IOT architectures are service centered. After surveying literature, there exists a gap in a user centered approach with intelligent edge devices. We propose Aura Minora: a novel architecture for IOT based on a bottom-up approach. Our approach in this paper is to see, how this user centric architecture would help end users as well as the smart city governing body to use existing Internet infrastructure to provide services. We also focus on user privacy and security aspect within this architecture. We further discuss the various views like informational, functional and deployment view of our architecture.

New Directions in the Teaching of Physical Sciences, 2017

Offering timely feedback on programming while encouraging learners to engage in critical evaluati... more Offering timely feedback on programming while encouraging learners to engage in critical evaluation of programs are the objectives of peer-testing. We report on a peer-testing experiment with students on distant campuses using a Web platform. The experiment shows the potential that peer-testing has to help students transition from passive learners to critical evaluators. Keywords: Computer science education, peer testing, peer feedback, software testing, student transitions

Computer Science & Information Technology ( CS & IT ), 2016

6th Annual International Conference on Computer Science Education: Innovation & Technology (CSEIT 2015), 2015

Lecture Notes in Networks and Systems

At crowded events, an increasing number of people use their mobile devices at the same time to ru... more At crowded events, an increasing number of people use their mobile devices at the same time to run applications having high data requirements leading to network congestion. This paper investigates the use of LED light to act as a Li-Fi source in order to transmit data. The data is captured using the CMOS camera of an Android smartphone. The experimental setup is tested using two different Android devices and the results are compared.

There are various definitions, view and explanations about Semantic Web, its usage and its underl... more There are various definitions, view and explanations about Semantic Web, its usage and its underlying architecture. However, the various flavours of explanations seem to have swayed way off-topic to the real purpose of Semantic Web. In this paper, we try to review the literature of Semantic Web based on the original views of the pioneers of Semantic Web which includes, Sir Tim Berners-Lee, Dean Allemang, Ora Lassila and James Hendler. Understanding the vision of the pioneers of any technology is cornerstone to the development. We have broken down Semantic Web into two approaches which allows us to reason with why Semantic Web is not mainstream.

Expert Systems, Aug 8, 2023

Expert systems are being extensively used to make critical decisions involving emotional analysis... more Expert systems are being extensively used to make critical decisions involving emotional analysis in affective computing. The evolution of deep learning algorithms has improved the potential for extracting value from multimodal emotional data. However, these black‐box algorithms do not often explain the heuristics behind processing the input features for achieving certain outputs. This study focuses on the risks of using black‐box deep learning models for critical tasks, such as emotion recognition, and describes how human understandable interpretations of the workings of these models are extremely important. This study utilizes one of the largest multimodal datasets available–CMU‐MOSEI. Many researchers have used the pre‐extracted features provided by the CMU Multimodal SDK with black‐box deep learning models making it difficult to interpret the contribution of its individual features. This study describes the implications of significant features from various modalities (audio, video, text) identified using XAI in Multimodal Emotion Recognition. It describes the process of curating reduced feature models by using the Gradient SHAP XAI method. These reduced models with highly contributing features achieve comparable and at times even better results compared to their corresponding all‐feature models as well as the baseline model GraphMFN. This study reveals that carefully selecting significant features for a model can help filter out irrelevant features, and attenuate the noise or bias caused by them, leading to an improved performance efficiency of the expert systems by making them transparent, easily interpretable, and trustworthy.

Lecture Notes in Networks and Systems

Lecture Notes in Networks and Systems

2018 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops), 2018

The rise of mobile phone usage across the world has given rise to many context aware applications... more The rise of mobile phone usage across the world has given rise to many context aware applications. To study the effects of context-aware application we have created a prototype application for the main demographics in a University. The prototype provides location based context-awareness in two facets: outside the university campus (outdoor) and inside the university campus (indoor). We have used Wi-Fi fingerprinting technology to achieve indoor positioning. Furthermore, the prototype is connected to Moodle, an online e-learning service via web services to provide course materials to the student based on the user, location (classroom) and time. The students can further interact with their peers spatially for discussions, as their location data is securely shared with their group mates. Our research evaluates the effect of context-awareness for two user groups; the students and the lecturers. The usage of both indoor and outdoor positioning within a single prototype, to provide the location aspect of context in our research, shows 94% acceptance rate between both the user groups.

2018 2nd International Conference on Smart Grid and Smart Cities (ICSGSC), 2018

Growing population, increasing urbanization and competition for shrinking resources, Food Securit... more Growing population, increasing urbanization and competition for shrinking resources, Food Security is a challenge around the world. The UN has made it a top priority in their Global Goals to promote sustainable agricultural practices with modern technologies and fair distribution system to sustain the world’s population. In this paper we investigate the implementation of a micro controlled environment agriculture system, the MicroCEA, a smart device that automates growing vegetables in an urban indoor residential setting. We have constructed and tested the device and the platform in the United Arab Emirates where about 80% of the food is imported.Our urban smart farming device is based on an extension of the MIT Food Computer version 1, which was adapted to the local environment based on the materials and equipment that could be locally sourced here. To develop our prototype the action research methodology was employed to leverage findings and to improve the next stage of the prototype development. We present our research into the design and testing of a prototyped Internet of Things (IoT) MicroCEA device after comparing other similar open source systems. A new IoT architecture for urban smart farming devices, a set of lessons learned that can be used in forwarding this field of study that can contribute toward achieving food security and improved nutrition and promote sustainable agriculture.

The relative influence of Semantic Web on Artificial Intelligence and vice versa is ever increasi... more The relative influence of Semantic Web on Artificial Intelligence and vice versa is ever increasing. However Artificial Intelligence and Semantic Web applications and research are going in different tangents as mentioned by the pioneers of Semantic Web. The real benefit would be on mixing semantic data for use in Artificially Intelligent apps so that reasoning can be done on open datasets for different entities. One way to do reasoning between entities would be by finding commonalities. that purpose we have created EnSense (Enhanced Sensing) — A web application that can find commonalities between two RDF objects using Semantic Web. We have devised an algorithm that could perform the task with greater speed, better number of relationships and quality as opposed to its closely related works. Our results have proven an 85% decrease in time taken and have 145% increase in the number of commonalities found. We provide a visual representation of the commonalities based on an interactive g...

Human emotion is one of the fundamental ways we express our feelings and currently the main focus... more Human emotion is one of the fundamental ways we express our feelings and currently the main focus is only facial expression for emotion extraction from an image. The aim of this paper is to enlighten the importance of other different features of an image in extracting the emotions apart from only facial features. The created system is able to accept an image uploaded by the user and detect the emotion individually. Firstly, the emotion is predicted based on the facial feature, then by the color and then by the text. Lastly, all three detected emotion is put to a comparison and a final output is displayed to the user. The created system went through an evaluation and the results were striking to see that the system prediction was often similar to the human prediction and the users were satisfied with the outcome.

2019 International Conference on Computational Intelligence and Knowledge Economy (ICCIKE), 2019

In recent years, data-driven (image based) methodologies like deep learning and computer vision h... more In recent years, data-driven (image based) methodologies like deep learning and computer vision have made computers immensely accurate in terms of identifying features inside images. Research in this area has given way to a relatively new set of deep learning models known as Generative models which generate images alongside identifying features inside them. These models, particularly conditional generative adversarial networks (CGANs), conditional variational autoencoders (CVAE) and generative stochastic networks (GSN) have become popular as they are able to translate images from one setting to another while keeping the structure of generated images aligned with the input images. In this paper, we review the work that has been done using these models to the area of web design automation which needs to be considered during the development phase. We also try to identify the benefits of implementing these models based on architectural features while keeping in view the different proble...

2017 International Conference on Computer and Applications (ICCA), 2017

Medical data contain very valuable information which can save many lives if it is analyzed and ut... more Medical data contain very valuable information which can save many lives if it is analyzed and utilized efficiently. Efficient analysis of this large volume of data demands the right choice of predictors and this in turn can impact the accuracy of the decision support system. Dimensionality reduction and feature subset selection are two techniques to reduce the number of features used in classification. In this paper we perform an empirical evaluation of four feature selection methods when applied in conjunction with Random Forest classifier. The feature selection techniques applied are Relief feature selection algorithm, Random forest selector, Recursive feature elimination and Boruta Feature selection algorithm. Results show that feature selection methods boosts the performance of the classifiers and in this case the features selected by the Boruta feature selection algorithm gives the best results.

Proceedings of the International Conference on Big Data and Advanced Wireless Technologies - BDAW '16, 2016

For a typical smart city many IoT (Internet of Things) architectures have been proposed to integr... more For a typical smart city many IoT (Internet of Things) architectures have been proposed to integrate the various ICT Solutions. Since there are many perspectives, the emphasis and the design of each architecture vary. So at the moment there is no standard architecture. The best approach is to adapt architecture for its intended use. Most of the current IOT architectures are service centered. After surveying literature, there exists a gap in a user centered approach with intelligent edge devices. We propose Aura Minora: a novel architecture for IOT based on a bottom-up approach. Our approach in this paper is to see, how this user centric architecture would help end users as well as the smart city governing body to use existing Internet infrastructure to provide services. We also focus on user privacy and security aspect within this architecture. We further discuss the various views like informational, functional and deployment view of our architecture.

New Directions in the Teaching of Physical Sciences, 2017

Offering timely feedback on programming while encouraging learners to engage in critical evaluati... more Offering timely feedback on programming while encouraging learners to engage in critical evaluation of programs are the objectives of peer-testing. We report on a peer-testing experiment with students on distant campuses using a Web platform. The experiment shows the potential that peer-testing has to help students transition from passive learners to critical evaluators. Keywords: Computer science education, peer testing, peer feedback, software testing, student transitions

Computer Science & Information Technology ( CS & IT ), 2016

6th Annual International Conference on Computer Science Education: Innovation & Technology (CSEIT 2015), 2015

Lecture Notes in Networks and Systems

At crowded events, an increasing number of people use their mobile devices at the same time to ru... more At crowded events, an increasing number of people use their mobile devices at the same time to run applications having high data requirements leading to network congestion. This paper investigates the use of LED light to act as a Li-Fi source in order to transmit data. The data is captured using the CMOS camera of an Android smartphone. The experimental setup is tested using two different Android devices and the results are compared.

There are various definitions, view and explanations about Semantic Web, its usage and its underl... more There are various definitions, view and explanations about Semantic Web, its usage and its underlying architecture. However, the various flavours of explanations seem to have swayed way off-topic to the real purpose of Semantic Web. In this paper, we try to review the literature of Semantic Web based on the original views of the pioneers of Semantic Web which includes, Sir Tim Berners-Lee, Dean Allemang, Ora Lassila and James Hendler. Understanding the vision of the pioneers of any technology is cornerstone to the development. We have broken down Semantic Web into two approaches which allows us to reason with why Semantic Web is not mainstream.