Search results for: in-context learning (ICL)

Authors: Huy Hieu Pham, Houssam Salmane, Louahdi Khoudour, Alain Crouzil, Pablo Zegers, Sergio Velastin

Abstract:

We present a deep learning-based multitask framework for joint 3D human pose estimation and action recognition from color video sequences. Our approach proceeds along two stages. In the first, we run a real-time 2D pose detector to determine the precise pixel location of important key points of the body. A two-stream neural network is then designed and trained to map detected 2D keypoints into 3D poses. In the second, we deploy the Efficient Neural Architecture Search (ENAS) algorithm to find an optimal network architecture that is used for modeling the Spatio-temporal evolution of the estimated 3D poses via an image-based intermediate representation and performing action recognition. Experiments on Human3.6M, Microsoft Research Redmond (MSR) Action3D, and Stony Brook University (SBU) Kinect Interaction datasets verify the effectiveness of the proposed method on the targeted tasks. Moreover, we show that our method requires a low computational budget for training and inference.

Keywords: human action recognition, pose estimation, D-CNN, deep learning

Procedia PDF Downloads 146

Authors: N. Pritchard

Abstract:

Within architectural education, students arrive fore-armed with; their life-experience; knowledge gained from subject-based learning; their brains and more specifically their imaginations. The learning-by-doing that they embark on in studio-based/project-based learning calls for supervision that allows the student to proactively undertake research and experimentation with design solution possibilities. The degree to which this supervision includes direction is subject to debate and differing opinion. It can be argued that if the student is to learn-by-doing, then design decision making within the design process needs to be instigated and owned by the student so that they have the ability to personally reflect on and evaluate those decisions. Within this premise lies the problem that the student's endeavours can become unstructured and unfocused as they work their way into a new and complex activity. A resultant weakness can be that the design activity is compartmented and not holistic or comprehensive, and therefore, the student's reflections are consequently impoverished in terms of providing a positive, informative feedback loop. The construct proffered in this paper is that a supportive 'armature' or 'Heuristic-Framework' can be developed that facilitates a holistic approach and reflective learning. The normal explorations of architectural design comprise: Analysing the site and context, reviewing building precedents, assimilating the briefing information. However, the student can still be compromised by 'not knowing what they need to know'. The long-serving triad 'Firmness, Commodity and Delight' provides a broad-brush framework of considerations to explore and integrate into good design. If this were further atomised in subdivision formed from the disparate aspects of architectural design that need to be considered within the design process, then the student could sieve through the facts more methodically and reflectively in terms of considering their interrelationship conflict and alliances. The words facts and sieve hold the acronym of the aspects that form the Heuristic-Framework: Function, Aesthetics, Context, Tectonics, Spatial, Servicing, Infrastructure, Environmental, Value and Ecological issues. The Heuristic could be used as a Hermeneutic Model with each aspect of design being focused on and considered in abstraction and then considered in its relation to other aspect and the design proposal as a whole. Importantly, the heuristic could be used as a method for gathering information and enhancing the design brief. The more poetic, mysterious, intuitive, unconscious processes should still be able to occur for the student. The Heuristic-Framework should not be seen as comprehensive prescriptive formulaic or inhibiting to the wide exploration of possibilities and solutions within the architectural design process.

Keywords: atomistic, hermeneutic, holistic, approach architectural design studio education

Procedia PDF Downloads 260

Authors: B. R. Campomanes-Alvarez, P. Quiros, B. Fernandez

Abstract:

Automatic Speech Recognition (ASR) is a machine-based process of decoding and transcribing oral speech. A typical ASR system receives acoustic input from a speaker or an audio file, analyzes it using algorithms, and produces an output in the form of a text. Some speech recognition systems use Hidden Markov Models (HMMs) to deal with the temporal variability of speech and Gaussian Mixture Models (GMMs) to determine how well each state of each HMM fits a short window of frames of coefficients that represents the acoustic input. Another way to evaluate the fit is to use a feed-forward neural network that takes several frames of coefficients as input and produces posterior probabilities over HMM states as output. Deep neural networks (DNNs) that have many hidden layers and are trained using new methods have been shown to outperform GMMs on a variety of speech recognition systems. Acoustic models for state-of-the-art ASR systems are usually training on massive amounts of data. However, audio files with their corresponding transcriptions can be difficult to obtain, especially in the Spanish language. Hence, in the case of these low-resource scenarios, building an ASR model is considered as a complex task due to the lack of labeled data, resulting in an under-trained system. Semi-supervised learning approaches arise as necessary tasks given the high cost of transcribing audio data. The main goal of this proposal is to develop a procedure based on acoustic semi-supervised learning for Spanish ASR systems by using DNNs. This semi-supervised learning approach consists of: (a) Training a seed ASR model with a DNN using a set of audios and their respective transcriptions. A DNN with a one-hidden-layer network was initialized; increasing the number of hidden layers in training, to a five. A refinement, which consisted of the weight matrix plus bias term and a Stochastic Gradient Descent (SGD) training were also performed. The objective function was the cross-entropy criterion. (b) Decoding/testing a set of unlabeled data with the obtained seed model. (c) Selecting a suitable subset of the validated data to retrain the seed model, thereby improving its performance on the target test set. To choose the most precise transcriptions, three confidence scores or metrics, regarding the lattice concept (based on the graph cost, the acoustic cost and a combination of both), was performed as selection technique. The performance of the ASR system will be calculated by means of the Word Error Rate (WER). The test dataset was renewed in order to extract the new transcriptions added to the training dataset. Some experiments were carried out in order to select the best ASR results. A comparison between a GMM-based model without retraining and the DNN proposed system was also made under the same conditions. Results showed that the semi-supervised ASR-model based on DNNs outperformed the GMM-model, in terms of WER, in all tested cases. The best result obtained an improvement of 6% relative WER. Hence, these promising results suggest that the proposed technique could be suitable for building ASR models in low-resource environments.

Keywords: automatic speech recognition, deep neural networks, machine learning, semi-supervised learning

Procedia PDF Downloads 339

Authors: Nagore Guerra Bilbao, Clemente Lobato Fraile

Abstract:

Mentoring is provided by professionals with a higher level of experience and competence as part of the professional development of a university faculty. This paper explores the characteristics of the mentoring provided by those teachers participating in the development of an active methodology program run at the University of the Basque Country: to examine and to analyze mentors’ performance with the aim of providing empirical evidence regarding its value as a lifelong learning strategy for teaching staff. A total of 183 teachers were trained during the first three programs. The analysis method uses a coding technique and is based on flexible, systematic guidelines for gathering and analyzing qualitative data. The results have confirmed the conception of mentoring as a methodological innovation in higher education. In short, university teachers in general assessed the mentoring they received positively, considering it to be a valid, useful strategy in their professional development. They highlighted the methodological expertise of their mentor and underscored how they monitored the learning process of the active method and provided guidance and advice when necessary. Finally, they also drew attention to traits such as availability, personal commitment and flexibility in. However, a minority critique is pointed to some aspects of the performance of some mentors.

Keywords: higher education, mentoring, professional development, university teachers

Procedia PDF Downloads 241

Authors: Md. Rasel Mia, Ashik Billah

Abstract:

The empowerment of girls and youth in Bangladesh is a demanding issue in today's digital age, where online learning and education have become integral to personal and societal development. This abstract explores the critical importance of creating a secure online environment for girls and youth in Bangladesh, emphasizing the transformative impact it can have on their access to education and knowledge. Bangladesh, like many developing nations, faces gender inequalities in education and access to digital resources. The creation of a safe digital space not only mitigates the gender digital divide but also fosters an environment where girls and youth can thrive academically and professionally. This manuscript draws attention to the efforts through a mixed-method study to assess the current digital landscape in Bangladesh, revealing disparities in phone and internet access, online practices, and awareness of cyber security among diverse demographic groups. Moreover, the study unveils the varying levels of familial support and barriers encountered by girls and youth in their quest for digital literacy. It emphasizes the need for tailored training programs that address specific learning needs while also advocating for enhanced internet accessibility, safe online practices, and inclusive online platforms. The manuscript culminates in a call for collaborative efforts among stakeholders, including NGOs, government agencies, and telecommunications companies, to implement targeted interventions that bridge the gender digital divide and pave the way for a brighter, more equitable future for girls and youth in Bangladesh. In conclusion, this research highlights the undeniable significance of creating a safe digital space as a catalyst for the empowerment of girls and youth in Bangladesh, ensuring that they not only access but excel in the online space, thereby contributing to their personal growth and the advancement of society as a whole.

Keywords: collaboration, cyber security, digital literacy, digital resources, inclusiveness

Procedia PDF Downloads 61

Authors: Seyed-Ali Sadegh-Zadeh, Mahboobe Bahrami, Sahar Ahmadi, Seyed-Yaser Mousavi, Hamed Atashbar, Amir M. Hajiyavand

Abstract:

This paper presents an investigation into the concept of neural reshaping, which is crucial for achieving strong artificial intelligence through the development of AI algorithms with very high plasticity. By examining the plasticity of both human and artificial neural networks, the study uncovers groundbreaking insights into how these systems adapt to new experiences and situations, ultimately highlighting the potential for creating advanced AI systems that closely mimic human intelligence. The uniqueness of this paper lies in its comprehensive analysis of the neural reshaping process in both human and artificial intelligence systems. This comparative approach enables a deeper understanding of the fundamental principles of neural plasticity, thus shedding light on the limitations and untapped potential of both human and AI learning capabilities. By emphasizing the importance of neural reshaping in the quest for strong AI, the study underscores the need for developing AI algorithms with exceptional adaptability and plasticity. The paper's findings have significant implications for the future of AI research and development. By identifying the core principles of neural reshaping, this research can guide the design of next-generation AI technologies that can enhance human and artificial intelligence alike. These advancements will be instrumental in creating a new era of AI systems with unparalleled capabilities, paving the way for improved decision-making, problem-solving, and overall cognitive performance. In conclusion, this paper makes a substantial contribution by investigating the concept of neural reshaping and its importance for achieving strong AI. Through its in-depth exploration of neural plasticity in both human and artificial neural networks, the study unveils vital insights that can inform the development of innovative AI technologies with high adaptability and potential for enhancing human and AI capabilities alike.

Keywords: neural plasticity, brain adaptation, artificial intelligence, learning, cognitive reshaping

Procedia PDF Downloads 52

Authors: Migena Ceyhan, Zeynep Orhan, Dimitrios Karras

Abstract:

People in modern societies are continuously sharing their experiences, emotions, and thoughts in different areas of life. The information reaches almost everyone in real-time and can have an important impact in shaping people’s way of living. This phenomenon is very well recognized and advantageously used by the market representatives, trying to earn the most from this means. Given the abundance of information, people and organizations are looking for efficient tools that filter the countless data into important information, ready to analyze. This paper is a modest contribution in this field, describing the process of automatically classifying social media comments in the Turkish language into positive or negative. Once data is gathered and preprocessed, feature sets of selected single words or groups of words are build according to the characteristics of language used in the texts. These features are used later to train, and test a system according to different machine learning algorithms (Naïve Bayes, Sequential Minimal Optimization, J48, and Bayesian Linear Regression). The resultant high accuracies can be important feedback for decision-makers to improve the business strategies accordingly.

Keywords: feature selection, machine learning, natural language processing, sentiment analysis, social media reviews

Procedia PDF Downloads 146

Authors: Panagiotis Kosmas

Abstract:

The advent of new technologies in education inspires practitioners to approach teaching from a different angle with the aim to professionally develop and improve teaching practices. Online communities of practice among teachers seem to be a trend associated with the integration efforts for a modern and pioneering educational system and training program. This study attempted to explore the participation in online communities of practice and the sharing of knowledge between teachers with aims to explore teachers' incentives to participate in such a community of practice. The study aims to contribute to international research, bringing in global debate new concerns and issues related to the professional learning of current educators. One official online community was used as a case study for the purposes of research. The data collection was conducted from the content analysis of online portal, by questionnaire in 184 community members and interviews with ten active users of the portal. The findings revealed that sharing of knowledge is a key motivation of members of a community. Also, the active learning and community participation seem to be essential factors for the success of an online community of practice.

Keywords: communities of practice, teachers, sharing knowledge, professional development

Procedia PDF Downloads 348

Authors: Morteza Pourazar, Fatemeh Mirakhori, Fazlolah Bagherzadeh, Rasool Hemayattalab

Abstract:

The purpose of study was to examine the effects of external and internal focus of attention in the motor learning of children with cerebral palsy. The study involved 30 boys (7 to 12 years old) with CP type 1 who practiced throwing beanbags. The participants were randomly assigned to the internal focus, external focus, and control groups, and performed six blocks of 10-trial with attentional focus reminders during a practice phase and no reminders during retention and transfer tests. Analysis of variance (ANOVA) with repeated measures on the last factor was used. The results show that significant main effects were found for time and group. However, the interaction of time and group was not significant. Retention scores were significantly higher for the external focus group. The external focus group performed better than other groups; however, the internal focus and control groups’ performance did not differ. The study concluded that motor skills in Spastic Hemiparetic Cerebral Palsy (SHCP) children could be enhanced by external attention.

Keywords: cerebral palsy, external attention, internal attention, throwing task

Procedia PDF Downloads 314

Authors: Al Romano, Jeryl C. Hije, Mechaela Marie O. Tabiolo

Abstract:

The decline in water quality is a major factor in aquatic disease outbreaks and can lead to significant mortality among aquatic organisms. Understanding the relationship between dissolved oxygen (DO) and water temperature is crucial, as these variables directly impact the health, behavior, and survival of fish populations. This study investigated how DO levels, water temperature, and atmospheric temperature interact in Sampaloc Lake to assess the risk of fish mortality. By employing a combination of linear regression models and machine learning techniques, researchers developed predictive models to forecast DO concentrations at various depths. The results indicate that while DO levels generally decrease with depth, the predicted concentrations are sufficient to support the survival of common fish species in Sampaloc Lake during March, April, and May 2025.

Keywords: aquaculture, dissolved oxygen, water temperature, regression analysis, machine learning, fish mass mortality, early warning system

Procedia PDF Downloads 35

Authors: Akaash Tawade, Manan Khattar, Lucas Spangher, Costas J. Spanos

Abstract:

Many grids are increasing the share of renewable energy in their generation mix, which is causing the energy generation to become less controllable. Buildings, which consume nearly 33% of all energy, are a key target for demand response: i.e., mechanisms for demand to meet supply. Understanding the behavior of office workers is a start towards developing demand response for one sector of building technology. The literature notes that dynamic computational modeling can be predictive of individual action, especially given that occupant behavior is traditionally abstracted from demand forecasting. Recent work founded on Social Cognitive Theory (SCT) has provided a promising conceptual basis for modeling behavior, personal states, and environment using control theoretic principles. Here, an adapted linear dynamical system of latent states and exogenous inputs is proposed to simulate energy demand amongst office workers engaged in a social energy shifting game. The energy shifting competition is implemented in an office in Singapore that is connected to a minigrid of buildings with a consistent 'price signal.' This signal is translated into a 'points signal' by a reinforcement learning (RL) algorithm to influence participant energy use. The dynamic model functions at the intersection of the points signals, baseline energy consumption trends, and SCT behavioral inputs to simulate future outcomes. This study endeavors to analyze how the dynamic model trains an RL agent and, subsequently, the degree of accuracy to which load deferability can be simulated. The results offer a generalizable behavioral model for energy competitions that provides the framework for further research on transfer learning for RL, and more broadly— transactive control.

Keywords: energy demand forecasting, social cognitive behavioral modeling, social game, transfer learning

Procedia PDF Downloads 107

Authors: Zachary Pan

Abstract:

Our general mission is to solve the problem of classifying images into different body part types and deciding if each of them is healthy or not. However, for now, we will determine healthiness for only one-sixth of the body parts, specifically the chest. We will detect pneumonia in X-ray scans of those chest images. With this type of AI, doctors can use it as a second opinion when they are taking CT or X-ray scans of their patients. Another ad-vantage of using this machine learning classifier is that it has no human weaknesses like fatigue. The overall ap-proach to this problem is to split the problem into two parts: first, classify the image, then determine if it is healthy. In order to classify the image into a specific body part class, the body parts dataset must be split into test and training sets. We can then use many models, like neural networks or logistic regression models, and fit them using the training set. Now, using the test set, we can obtain a realistic accuracy the models will have on images in the real world since these testing images have never been seen by the models before. In order to increase this testing accuracy, we can also apply many complex algorithms to the models, like multiplicative weight update. For the second part of the problem, to determine if the body part is healthy, we can have another dataset consisting of healthy and non-healthy images of the specific body part and once again split that into the test and training sets. We then use another neural network to train on those training set images and use the testing set to figure out its accuracy. We will do this process only for the chest images. A major conclusion reached is that convolutional neural networks are the most reliable and accurate at image classification. In classifying the images, the logistic regression model, the neural network, neural networks with multiplicative weight update, neural networks with the black box algorithm, and the convolutional neural network achieved 96.83 percent accuracy, 97.33 percent accuracy, 97.83 percent accuracy, 96.67 percent accuracy, and 98.83 percent accuracy, respectively. On the other hand, the overall accuracy of the model that de-termines if the images are healthy or not is around 78.37 percent accuracy.

Keywords: body part, healthcare, machine learning, neural networks

Procedia PDF Downloads 103

Authors: Pimpisa Rattanadilok Na Phuket

Abstract:

This research study aims to examine: 1) teaching competency of the native English-speaking teacher (NEST) 2) the English language learning achievement of Thai students, and 3) students’ perceptions toward their NEST. The population considered in this research was a group of 39 undergraduate students of the academic year 2013. The tools consisted of a questionnaire employed to measure the level of competency of NEST, pre-test and post-test used to examine the students’ achievement on English pronunciation, and an interview used to discover how participants perceived their NEST. The data was statistically analysed as percentage, mean, standard deviation and One-sample-t-test. In addition, the data collected by interviews was qualitatively analyzed. The research study found that the level of teaching competency of native speaker teachers of English was mostly low, the English pronunciation achievement of students had increased significantly at the level of 0.5, and the students’ perception toward NEST is combined. The students perceived their NEST as an English expertise, but they felt that NEST had not recognized students' linguistic difficulty and cultural differences.

Keywords: competency, native English-speaking teacher (NET), English teaching, learning achievement

Procedia PDF Downloads 374

Authors: Chan Ka Lok Sobel

Abstract:

The rapid evolution of digital media platforms has given rise to new forms of narrative engagement, particularly through multi-channel storytelling. This research focuses on exploring the teaching pedagogy and effectiveness of multi-channel storytelling for social media, cinema, games, and streaming platforms. The study employs case studies of the popular series "Squid Game" to investigate the diverse pedagogical approaches and strategies used in teaching multi-channel storytelling. Through qualitative research methods, including interviews, surveys, and content analysis, the research assesses the effectiveness of these approaches in terms of student engagement, knowledge acquisition, critical thinking skills, and the development of digital literacy. The findings contribute to understanding best practices for incorporating multi-channel storytelling into educational contexts and enhancing learning outcomes in the digital media landscape.

Keywords: digital literacy, game-based learning, artificial intelligence, animation production, educational technology

Procedia PDF Downloads 114

Authors: Selim M. Khan, Aaron A. Goodarzi, Joshua M. Taron, Tryggve Rönnqvist

Abstract:

Indoor air quality, a prime determinant of health, is strongly influenced by the presence of hazardous radon gas within the built environment. As a health issue, dangerously high indoor radon arose within the 20th century to become the 2nd leading cause of lung cancer. While the 21st century building metrics and human behaviors have captured, contained, and concentrated radon to yet higher and more hazardous levels, the issue is rapidly worsening in Canada. It is established that Canadians in the Prairies are the 2nd highest radon-exposed population in the world, with 1 in 6 residences experiencing 0.2-6.5 millisieverts (mSv) radiation per week, whereas the Canadian Nuclear Safety Commission sets maximum 5-year occupational limits for atomic workplace exposure at only 20 mSv. This situation is also deteriorating over time within newer housing stocks containing higher levels of radon. Deep machine learning (LSTM) algorithms were applied to analyze multiple quantitative and qualitative features, determine the most important contributory factors, and predicted radon levels in the known past (1990-2020) and projected future (2021-2050). The findings showed gradual downwards patterns in Sweden, whereas it would continue to go from high to higher levels in Canada over time. The contributory factors found to be the basement porosity, roof insulation depthness, R-factor, and air dynamics of the indoor environment related to human window opening behaviour. Building codes must consider including these factors to ensure adequate indoor ventilation and healthy living that can prevent lung cancer in non-smokers.

Keywords: radon, building metrics, deep learning, LSTM prediction model, lung cancer, canada, sweden

Procedia PDF Downloads 112

Authors: Nádia Pereira, Paula Ferreira, Sofia Francisco, Sofia Oliveira, Sidclay Souza, Paula Paulino, Ana Margarida Veiga Simão

Abstract:

Social media has progressed into a platform for hate speech among its users, and thus, there is an increasing need to develop automatic detection classifiers of offense and conflicts to help decrease the prevalence of such incidents. Online communication can be used to intentionally harm someone, which is why such classifiers could be essential in social networks. A possible application of these classifiers is the automatic detection of cyberbullying. Even though identifying the aggressive language used in online interactions could be important to build cyberbullying datasets, there are other criteria that must be considered. Being able to capture the language, which is indicative of the intent to harm others in a specific context of online interaction is fundamental. Offense and hate speech may be the foundation of online conflicts, which have become commonly used in social media and are an emergent research focus in machine learning and natural language processing. This study presents two Portuguese language offense-related datasets which serve as examples for future research and extend the study of the topic. The first is similar to other offense detection related datasets and is entitled Aggressiveness dataset. The second is a novelty because of the use of the history of the interaction between users and is entitled the Conflicts/Attacks dataset. Both datasets were developed in different phases. Firstly, we performed a content analysis of verbal aggression witnessed by adolescents in situations of cyberbullying. Secondly, we computed frequency analyses from the previous phase to gather lexical and linguistic cues used to identify potentially aggressive conflicts and attacks which were posted on Twitter. Thirdly, thorough annotation of real tweets was performed byindependent postgraduate educational psychologists with experience in cyberbullying research. Lastly, we benchmarked these datasets with other machine learning classifiers.

Keywords: aggression, classifiers, cyberbullying, datasets, hate speech, machine learning

Procedia PDF Downloads 228

Authors: Muzaffar Hussain

Abstract:

Requests strategies are considered as a part of the speech acts, which are frequently used in everyday communication. Each language provides speech acts to the speakers; therefore, the selection of appropriate form seems more culture-specific rather than language. The present paper investigates the gender-based difference in the use of request strategies by native speakers of Urdu/Punjabi male and female who are learning English as a second language. The data for the present study were collected from 68 graduate students, who are learning English as an L2 in Pakistan. They were given an online close-ended questionnaire, based on Discourse Completion Test (DCT). After analyzing the data, it was found that the L1 male Urdu/Punjabi speakers were inclined to use more direct request strategies while the female Urdu/Punjabi speakers used indirect request strategies. This paper also found that in some situations female participants used more direct strategies than male participants. The present study concludes that the use of request strategies is influenced by culture, social status, and power distribution in a society.

Keywords: gender variation, request strategies, face-threatening, second language pragmatics, language competence

Procedia PDF Downloads 189

Authors: Kamaludeen Samaila, Khashyaullah Abdulfattah, Fahimi Ahmad Bin Amir

Abstract:

The study was conducted to examine the relationship between selected factors (performance expectancy, effort expectancy, social influence and facilitating condition) and students’ intention to use the learning management system (LMS), as well as investigating the factors predicting students’ intention to use the LMS. The study was specifically conducted at the Faculty of Educational Study of University Putra Malaysia. Questionnaires were distributed to 277 respondents using a random sampling technique. SPSS Version 22 was employed in analyzing the data; the findings of this study indicated that performance expectancy (r = .69, p < .01), effort expectancy (r=.60, p < .01), social influence (r = .61, p < .01), and facilitating condition (r=.42, p < .01), were significantly related to students’ intention to use the LMS. In addition, the result also revealed that performance expectancy (β = .436, p < .05), social influence (β=.232, p < .05), and effort expectancy (β = .193, p < .05) were strong predictors of students’ intention to use the LMS. The analysis further indicated that (R2) is 0.054 which means that 54% of variation in the dependent variable is explained by the entire predictor variables entered into the regression model. Understanding the factors that affect students’ intention to use the LMS could help the lecturers, LMS managers and university management to develop the policies that may attract students to use the LMS.

Keywords: LMS, postgraduate students, PutraBlas, students’ intention, UPM, UTAUT model

Procedia PDF Downloads 510

Authors: Joshua van Staden, Dane Brown, Karen Bradshaw

Abstract:

Gun violence is a significant problem in modern-day society. Early detection of carried handguns through closed-circuit television (CCTV) can aid in preventing potential gun violence. However, CCTV operators have a limited attention span. Machine learning approaches to automating the detection of dangerous gun carriers provide a way to aid CCTV operators in identifying these individuals. This study provides insight into the relationship between human key points extracted using human pose estimation (HPE) and their intention to fire a weapon. We examine the feature importance of each keypoint and their correlations. We use principal component analysis (PCA) to reduce the feature space and optimize detection. Finally, we run a set of classifiers to determine what form of classifier performs well on this data. We find that hips, shoulders, and knees tend to be crucial aspects of the human pose when making these predictions. Furthermore, the horizontal position plays a larger role than the vertical position. Of the 66 key points, nine principal components could be used to make nonlinear classifications with 86% accuracy. Furthermore, linear classifications could be done with 85% accuracy, showing that there is a degree of linearity in the data.

Keywords: feature engineering, human pose, machine learning, security

Procedia PDF Downloads 93

Authors: Zhi Liu, Xian Peng, Monika Domanska, Lingyun Kang, Sannyuya Liu

Abstract:

Massive Open education has become increasingly popular among worldwide learners. An increasing number of course reviews are being generated in Massive Open Online Course (MOOC) platform, which offers an interactive feedback channel for learners to express opinions and feelings in learning. These reviews typically contain subjective emotion and topic information towards the courses. However, it is time-consuming to artificially detect these opinions. In this paper, we propose an emotion-oriented topic detection model to automatically detect the students’ opinioned aspects in course reviews. The known overall emotion orientation and emotional words in each review are used to guide the joint probabilistic modeling of emotion and aspects in reviews. Through the experiment on real-life review data, it is verified that the distribution of course-emotion-aspect can be calculated to capture the most significant opinioned topics in each course unit. This proposed technique helps in conducting intelligent learning analytics for teachers to improve pedagogies and for developers to promote user experiences.

Keywords: Massive Open Online Course (MOOC), course reviews, topic model, emotion recognition, topical aspects

Procedia PDF Downloads 262

Authors: Elham A. Alsadoon, Hamadah B. Alsadoon

Abstract:

Educators try to design learning environments that are preferred by their students. With the wide-spread adoption of cell phones surpassing any other technology, educators should not fail to invest in the power of such technology. This study aimed to explore the current use of cell phones in education among Saudi students in Saudi universities and how students perceive such use. Data was collected from 237 students at King Saud University. Descriptive analysis was used to analyze the data. A T-test for independent groups was used to examine whether there was a significant difference between males and females in their perception of using cell phones in education. Findings suggested that students have a positive attitude toward the use of cell phones in education. The most accepted use was for sending notification to students, which has already been experienced through the Twasel system provided by King Saud University. This electronic system allows instructors to easily send any SMS or email to their students. The use of cell phone applications came in the second rank of using cell phones in education. Students have already experienced the benefits of having these applications handy wherever they go. On the other hand, they did not perceive using cell phones for assessment as practical educational usage. No gender difference was detected in terms of students’ perceptions toward using cell phones in education.

Keywords: cell phone, mobile learning, educational sciences, education

Procedia PDF Downloads 413

Authors: Khairani Binti Supyan, Fatimah Khalid, Mas Rina Mustaffa, Azreen Bin Azman, Amirul Azuani Romle

Abstract:

Perennial plant classification plays a significant role in various agricultural and environmental applications, assisting in plant identification, disease detection, and biodiversity monitoring. Nevertheless, attaining high accuracy in perennial plant image classification remains challenging due to the complex variations in plant appearance, the diverse range of environmental conditions under which images are captured, and the inherent variability in image quality stemming from various factors such as lighting conditions, camera settings, and focus. This paper proposes an adaptation approach to optimize perennial plant image classification by fine-tuning the pre-trained DNNs model. This paper explores the efficacy of fine-tuning prevalent architectures, namely VGG16, ResNet50, and InceptionV3, leveraging transfer learning to tailor the models to the specific characteristics of perennial plant datasets. A subset of the MYLPHerbs dataset consisted of 6 perennial plant species of 13481 images under various environmental conditions that were used in the experiments. Different strategies for fine-tuning, including adjusting learning rates, training set sizes, data augmentation, and architectural modifications, were investigated. The experimental outcomes underscore the effectiveness of fine-tuning deep neural networks for perennial plant image classification, with ResNet50 showcasing the highest accuracy of 99.78%. Despite ResNet50's superior performance, both VGG16 and InceptionV3 achieved commendable accuracy of 99.67% and 99.37%, respectively. The overall outcomes reaffirm the robustness of the fine-tuning approach across different deep neural network architectures, offering insights into strategies for optimizing model performance in the domain of perennial plant image classification.

Keywords: perennial plants, image classification, deep neural networks, fine-tuning, transfer learning, VGG16, ResNet50, InceptionV3

Procedia PDF Downloads 66

Authors: Maurice Ntahobari, Levin Kuhlmann, Mario Boley, Zhinoos Razavi Hesabi

Abstract:

For machine learning based epileptic seizure prediction, it is important for the model to be implemented in small implantable or wearable devices that can be used to monitor epilepsy patients; however, current state-of-the-art methods are complex and computationally intensive. We use Shapley Additive Explanation (SHAP) to find relevant intracranial electroencephalogram (iEEG) features and improve the computational efficiency of a state-of-the-art seizure prediction method based on the extra trees classifier while maintaining prediction performance. Results for a small contest dataset and a much larger dataset with continuous recordings of up to 3 years per patient from 15 patients yield better than chance prediction performance (p < 0.004). Moreover, while the performance of the SHAP-based model is comparable to that of the benchmark, the overall training and prediction time of the model has been reduced by a factor of 1.83. It can also be noted that the feature called zero crossing value is the best EEG feature for seizure prediction. These results suggest state-of-the-art seizure prediction performance can be achieved using efficient methods based on optimal feature selection.

Keywords: machine learning, seizure prediction, extra tree classifier, SHAP, epilepsy

Procedia PDF Downloads 113

Authors: N. D'Amico, E. Grossi, B. Colombo, F. Rigiroli, M. Buscema, D. Fazzini, G. Cornalba, S. Papa

Abstract:

Purpose: To characterize, through a radiomic approach, the nature of nodules considered non-specific by expert radiologists, recognized in magnetic resonance mammography (MRm) with T1-weighted (T1w) sequences with paramagnetic contrast. Material and Methods: 47 cases out of 1200 undergoing MRm, in which the MRm assessment gave uncertain classification (non-specific nodules), were admitted to the study. The clinical outcome of the non-specific nodules was later found through follow-up or further exams (biopsy), finding 35 benign and 12 malignant. All MR Images were acquired at 1.5T, a first basal T1w sequence and then four T1w acquisitions after the paramagnetic contrast injection. After a manual segmentation of the lesions, done by a radiologist, and the extraction of 150 radiomic features (30 features per 5 subsequent times) a machine learning (ML) approach was used. An evolutionary algorithm (TWIST system based on KNN algorithm) was used to subdivide the dataset into training and validation test and to select features yielding the maximal amount of information. After this pre-processing, different machine learning systems were applied to develop a predictive model based on a training-testing crossover procedure. 10 cases with a benign nodule (follow-up older than 5 years) and 18 with an evident malignant tumor (clear malignant histological exam) were added to the dataset in order to allow the ML system to better learn from data. Results: NaiveBayes algorithm working on 79 features selected by a TWIST system, resulted to be the best performing ML system with a sensitivity of 96% and a specificity of 78% and a global accuracy of 87% (average values of two training-testing procedures ab-ba). The results showed that in the subset of 47 non-specific nodules, the algorithm predicted the outcome of 45 nodules which an expert radiologist could not identify. Conclusion: In this pilot study we identified a radiomic approach allowing ML systems to perform well in the diagnosis of a non-specific nodule at MR mammography. This algorithm could be a great support for the early diagnosis of malignant breast tumor, in the event the radiologist is not able to identify the kind of lesion and reduces the necessity for long follow-up. Clinical Relevance: This machine learning algorithm could be essential to support the radiologist in early diagnosis of non-specific nodules, in order to avoid strenuous follow-up and painful biopsy for the patient.

Keywords: breast, machine learning, MRI, radiomics

Procedia PDF Downloads 267

Authors: Guru Prakash, Revanth Dugalam

Abstract:

A road’s condition often deteriorates due to harsh loading such as overload due to trucks, and severe environmental conditions such as heavy rain, snow load, and cyclic loading. In absence of proper maintenance planning, this results in potholes, wide cracks, bumps, and increased roughness of roads. In this paper, a data-driven model will be developed to detect these damages using vibration and image signals. The key idea of the proposed methodology is that the road anomaly manifests in these signals, which can be detected by training a machine learning algorithm. The use of various machine learning techniques such as the support vector machine and Radom Forest method will be investigated. The proposed model will first be trained and tested with artificially simulated data, and the model architecture will be finalized by comparing the accuracies of various models. Once a model is fixed, the field study will be performed, and data will be collected. The field data will be used to validate the proposed model and to predict the future road’s health condition. The proposed will help to automate the road condition monitoring process, repair cost estimation, and maintenance planning process.

Keywords: SVM, data-driven, road health monitoring, pot-hole

Procedia PDF Downloads 86

Authors: Lisa Graham, Kathleen Grant

Abstract:

South Africa is a multicultural country with eleven official languages, yet most of the formal education at institutions of higher education in the country is in English. It is well known that many students, irrespective of their home language, struggle to grasp difficult scientific concepts and the same is true for students enrolled in the Extended Curriculum Programme at the Cape Peninsula University of Technology (CPUT), studying biomedical sciences. Today we are fortunate in that there is a plethora of resources available to students to research and better understand subject matter online. For example, the students often use YouTube videos to supplement the formal education provided in our course. Unfortunately, most of this material is presented in English. The rationale behind this project lies in that it is well documented that students think and grasp concepts easier in their home language and addresses the fact that the lingua franca of instruction in the field of biomedical science is English. A project aimed at addressing the lack of available resources in most of the South African languages is planned, where students studying Bachelor of Health Science in Medical Laboratory Science will collaborate with those studying Film and Video Technology to create educational videos, explaining scientific concepts in their home languages. These videos will then be published on our own YouTube channel, thereby making them accessible to fellow students, future students and anybody with interest in the subject. Research will be conducted to determine the benefit of the project as well as the published videos to the student community. It is suspected that the students engaged in making the videos will benefit in such a way as to gain further understanding of their course content, a broader appreciation of the discipline, an enhanced sense of civic responsibility, as well as greater respect for the different languages and cultures in our classes. Indeed, an increase in student engagement has been shown to play a central role in student success, and it is well noted that deeper learning and more innovative solutions take place in collaborative groups. We aim to make a meaningful contribution towards the production and repository of knowledge in multilingual teaching and learning for the benefit of the diverse student population and staff. This would strengthen language development, multilingualism, and multiculturalism at CPUT and empower and promote African languages as languages of science and education at CPUT, in other institutions of higher learning, and in South Africa as a whole.

Keywords: educational videos, multiculturalism, multilingualism, student engagement

Procedia PDF Downloads 155

Authors: Mehrnaz Mostafavi, Mahtab Shabani, Alireza Azani, Fatemeh Ghafari

Abstract:

Artificial intelligence (AI) can transform breast cancer diagnosis and therapy by providing sophisticated solutions for screening, imaging interpretation, histopathological analysis, and treatment planning. This literature review digs into the many uses of AI in breast cancer treatment, highlighting the need for collaboration between AI scientists and healthcare practitioners. It emphasizes advances in AI-driven breast imaging interpretation, such as computer-aided detection and diagnosis (CADe/CADx) systems and deep learning algorithms. These have shown significant potential for improving diagnostic accuracy and lowering radiologists' workloads. Furthermore, AI approaches such as deep learning have been used in histopathological research to accurately predict hormone receptor status and categorize tumor-associated stroma from regular H&E stains. These AI-powered approaches simplify diagnostic procedures while providing insights into tumor biology and prognosis. As AI becomes more embedded in breast cancer care, it is crucial to ensure its ethical, efficient, and patient-focused implementation to improve outcomes for breast cancer patients ultimately.

Keywords: breast cancer, artificial intelligence, cancer diagnosis, clinical practice

Procedia PDF Downloads 69

Authors: Mirza Waseem Abbas, Syed Danish Raza

Abstract:

For many countries agriculture is still the major force driving the economy and a critically important socioeconomic sector, despite exceptional industrial development across the globe. In countries like Pakistan, this sector is considered the backbone of the economy, and most of the economic decision making revolves around agricultural outputs and data. Timely and accurate facts and figures about this vital sector hold immense significance and have serious implications for the long-term development of the economy. Therefore, any significant improvements in the statistics and other forms of data regarding agriculture sector are considered important by all policymakers. This is especially true for decision making for the betterment of crops and the agriculture sector in general. Provincial and federal agricultural departments collect data for all cash and non-cash crops and the sector, in general, every year. Traditional data collection for such a large sector i.e. agriculture, being time-consuming, prone to human error and labor-intensive, is slowly but gradually being replaced by remote sensing techniques. For this study, remotely sensed data were used for change detection (machine learning, supervised & unsupervised classification) to assess the increase or decrease in area under agriculture over the last fifteen years due to urbanization. Detailed Landsat Images for the selected agricultural districts were acquired for the year 2000 and compared to images of the same area acquired for the year 2016. Observed differences validated through detailed analysis of the areas show that there was a considerable decrease in vegetation during the last fifteen years in four major agricultural districts of the Punjab province due to urbanization (housing societies).

Keywords: change detection, area estimation, machine learning, urbanization, remote sensing

Procedia PDF Downloads 249

Authors: Andreas Johansson

Abstract:

Conceptual understanding needs to be grounded on observation of physical phenomena, experiences or metaphors. Observation of physical phenomena using demonstration experiments has a long tradition within physics education and students need to develop mental models to relate the observations to concepts from scientific theories. This study investigates how live and video experiments involving an acoustic trap to visualize particle-field interaction, field properties and particle properties can help develop students' mental models and how they can be used differently to realize their potential as teaching tools. Initially, they were treated as analogs and the lesson designs were kept identical. With a design-based approach, the experimental and video designs, as well as best practices for a respective teaching tool, were then developed in iterations. Variation theory was used as a theoretical framework to analyze the planned respective realized pattern of variation and invariance in order to explain learning outcomes as measured by a pre-posttest consisting of conceptual multiple-choice questions inspired by the Force Concept Inventory and the Force and Motion Conceptual Evaluation. Interviews with students and teachers were used to inform the design of experiments and videos in each iteration. The lesson designs and the live and video experiments has been developed to help teachers improve student learning and make school physics more interesting by involving experimental setups that usually are out of reach and to bridge the gap between what happens in classrooms and in science research. As students’ conceptual knowledge also rises their interest in physics the aim is to increase their chances of pursuing careers within science, technology, engineering or mathematics.

Keywords: acoustic trap, design-based research, experiments, variation theory

Procedia PDF Downloads 83

Authors: William Whiteley, Jens Gregor

Abstract:

In recent developments of direct neural network based positron emission tomography (PET) reconstruction, two prominent architectures have emerged for converting measurement data into images: 1) networks that contain fully-connected layers; and 2) networks that primarily use a convolutional encoder-decoder architecture. In this paper, we present a multi-modal direct PET reconstruction method called MDPET, which is a hybrid approach that combines the advantages of both types of networks. MDPET processes raw data in the form of sinograms and histo-images in concert with attenuation maps to produce high quality multi-slice PET images (e.g., 8x440x440). MDPET is trained on a large whole-body patient data set and evaluated both quantitatively and qualitatively against target images reconstructed with the standard PET reconstruction benchmark of iterative ordered subsets expectation maximization. The results show that MDPET outperforms the best previously published direct neural network methods in measures of bias, signal-to-noise ratio, mean absolute error, and structural similarity.

Keywords: deep learning, image reconstruction, machine learning, neural network, positron emission tomography

Procedia PDF Downloads 111