Search results for: deep graphical model

Authors: Bhuyain Mobarok Hossain

Abstract:

Image classification and localization research is currently an important strategy in the field of computer vision. The evolution and advancement of deep learning and convolutional neural networks (CNN) have greatly improved the capabilities of object detection and image-based classification. Target detection is important to research in the field of computer vision, especially in video surveillance systems. To solve this problem, we will be applying a convolutional neural network of multiple scales at multiple locations in the image in one sliding window. Most translation networks move away from the bounding box around the area of interest. In contrast to this architecture, we consider the problem to be a classification problem where each pixel of the image is a separate section. Image classification is the method of predicting an individual category or specifying by a shoal of data points. Image classification is a part of the classification problem, including any labels throughout the image. The image can be classified as a day or night shot. Or, likewise, images of cars and motorbikes will be automatically placed in their collection. The deep learning of image classification generally includes convolutional layers; the invention of it is referred to as a convolutional neural network (CNN).

Keywords: image classification, object detection, localization, particle filter

Procedia PDF Downloads 305

Authors: Jerome D. Donovan

Abstract:

The pandemic has resulted in a dramatic re-consideration of the reliance on international student fees to support university models in Australia. A key resulting initiative for the Australian Federal Government has been shifting the way universities consider their research model, emphasising the importance of commercialising research. This study specifically examines this shift from the perspective of a university spin-off, examining how university support structures and incubation models have assisted in the translation of fundamental research into a high-growth university spin-off. A focused case study approach is adopted in this study, using an auto-ethnographic research method to document the experiences and insights drawn from being a co-founder in a university spin-off in a time where research commercialisation has emerged as a central focus in Australian universities.

Keywords: research commercialisation, spin-offs, university incubation, entrepreneurship

Procedia PDF Downloads 81

Authors: Shaibu Baanni Azumah, William Adzawla

Abstract:

Rice is an important staple crop, with current demand higher than the domestic supply in Ghana. This has led to a high and unfavourable import bill. Therefore, recent policies and interventions in the agricultural sub-sector aim at promoting various improved agricultural technologies in order to improve domestic production and reduce the importation of rice. In this study, we examined the effect of the adoption of Urea Deep Placement (UDP) technology by rice farmers on the position of the production frontier. This involved 200 farmers selected through a multi stage sampling technique in the Northern region of Ghana. A Cobb-Douglas stochastic frontier model was fitted. The result showed that the adoption of UDP technology shifts the output frontier outward and also move the farmers closer to the frontier. Farmers were also operating under diminishing returns to scale which calls for redress. Other factors that significantly influenced rice production were farm size, labour, use of certified seeds and NPK fertilizer. Although there was an opportunity for improvement, the farmers were highly efficient (92%), compared to previous studies. Farmers’ efficiency was improved through increased education, household size, experience, access to credit, and lack of extension service provision by MoFA. The study recommends the revision of Ghana’s agricultural policy to include the UDP technology. Agricultural Extension officers of the Ministry of Food and Agriculture (MoFA) should be trained on the UDP technology to support IFDC’s drive to improve adoption by rice farmers. Rice farmers are also encouraged to expand their farm lands, improve plant population, and also increase the usage of fertilizer to improve yields. Mechanisms through which credit can be made easily accessible and effectively utilised should be identified and promoted.

Keywords: efficiency, rice farmers, stochastic frontier, UDP technology

Procedia PDF Downloads 409

Authors: Lu Zhao, Nadir Farhi, Yeltsin Valero, Zoi Christoforou, Nadia Haddadou

Abstract:

In this paper, a long short-term memory (LSTM) neural network model is proposed to replicate simultaneously car-following and lane-changing behaviors in road networks. By combining two kinds of LSTM layers and three input designs of the neural network, six variants of the LSTM model have been created. These models were trained and tested on the NGSIM 101 dataset, and the results were evaluated in terms of longitudinal speed and lateral position, respectively. Then, we compared the LSTM model with a classical car-following model (the intelligent driving model (IDM)) in the part of speed decision. In addition, the LSTM model is compared with a model using classical neural networks. After the comparison, the LSTM model demonstrates higher accuracy than the physical model IDM in terms of car-following behavior and displays better performance with regard to both car-following and lane-changing behavior compared to the classical neural network model.

Keywords: traffic modeling, neural networks, LSTM, car-following, lane-change

Procedia PDF Downloads 261

Authors: Vera G. Dianova, Lori P. Montross, Charles M. Burke

Abstract:

In the digital age, the widespread and frequently excessive use of mobile phones amongst university students is recognized as a significant distractor which interferes with their ability to enter a deep state of concentration during studies and diminishes their prospects of experiencing the enjoyable and instrumental state of flow, as defined and described by psychologist M. Csikszentmihalyi. This study has targeted 50 university students with the aim of teaching them to cultivate their ability to engage in deep work and to attain the state of flow, fostering more effective and enjoyable learning experiences. Prior to the start of the intervention, all participating students completed a comprehensive survey based on a variety of validated scales assessing their inclination toward lifelong learning, frequency of flow experiences during study, frustration tolerance, sense of agency, as well as their love of learning and daily time devoted to non-academic mobile phone activities. Several days after this initial assessment, students received a 90-minute lecture on the principles of flow and deep work, accompanied by a critical discourse on the detrimental effects of excessive mobile phone usage. They were encouraged to practice deep work and strive for frequent flow states throughout the semester. Subsequently, students submitted weekly surveys, including the 10-item CORE Dispositional Flow Scale, a 3-item agency scale and furthermore disclosed their average daily hours spent on non-academic mobile phone usage. As a final step, at the end of the semester students engaged in reflective report writing, sharing their experiences and evaluating the intervention's effectiveness. They considered alterations in their love of learning, reflected on the implications of their mobile phone usage, contemplated improvements in their tolerance for boredom and perseverance in complex tasks, and pondered the concept of lifelong learning. Additionally, students assessed whether they actively took steps towards managing their recreational phone usage and towards improving their commitment to becoming lifelong learners. Employing a mixed-methods approach our study offers insights into the dynamics of concentration, flow, mobile phone usage and attitudes towards learning among undergraduate and graduate university students. The findings of this study aim to promote profound contemplation, on the part of both students and instructors, on the rapidly evolving digital-age higher education environment. In an era defined by digital and AI advancements, the ability to concentrate, to experience the state of flow, and to love learning has never been more crucial. This study underscores the significance of addressing mobile phone distractions and providing strategies for cultivating deep concentration. The insights gained can guide educators in shaping effective learning strategies for the digital age. By nurturing a love for learning and encouraging lifelong learning, educational institutions can better prepare students for a rapidly changing labor market, where adaptability and continuous learning are paramount for success in a dynamic career landscape.

Keywords: deep work, flow, higher education, lifelong learning, love of learning

Procedia PDF Downloads 68

Authors: Coskun Serefoglu

Abstract:

The study aims to analyse how a traditional sector such as agri-food could be mobilized through regional innovation strategies. A principal component analysis as well as qualitative information, such as in-depth interviews, focus group and surveys, were employed to find the priority sectors. An agri-food model was developed which includes both a linear model and interactive model. The model consists of two main components, one of which is technological integration and the other one is agricultural extension which is based on Land-grant university approach of U.S. which is not a common practice in Turkey.

Keywords: regional innovation strategy, interactive model, agri-food sector, local development, planning, regional development

Procedia PDF Downloads 149

Authors: Sasiporn Rattanasupha, Settapat Chinviriyasit

Abstract:

The treatment function adopts a continuous and differentiable function which can describe the effect of delayed treatment when the number of infected individuals increases and the medical condition is limited. In this paper, the SEIR epidemic model with treatment function is studied to investigate the dynamics of the model due to the effect of treatment. It is assumed that the treatment rate is proportional to the number of infective patients. The stability of the model is analyzed. The model is simulated to illustrate the analytical results and to investigate the effects of treatment on the spread of infection.

Keywords: basic reproduction number, local stability, SEIR epidemic model, treatment function

Procedia PDF Downloads 521

Authors: Wilson Enríquez, Daniel Cardenas

Abstract:

This article proposes a filter bank format combined with the mathematical tool called polyphase decomposition and the discrete Fourier transform (DFT) with the purpose of improving the performance of the fifth-generation communication systems (5G). We started with a review of the literature and the study of the filter bank theory and its combination with DFT in order to improve the performance of wireless communications since it reduces the computational complexity of these communication systems. With the proposed technique, several experiments were carried out in order to evaluate the structures in 5G systems. Finally, the results are presented in graphical form in terms of bit error rate against the ratio bit energy/noise power spectral density (BER vs. Eb / No).

Keywords: multi-carrier system (5G), filter bank, polyphase decomposition, FIR equalizer

Procedia PDF Downloads 201

Authors: Kevin Wang, Jian Ge, Yinan Zhao, Kevin Willis

Abstract:

Kepler has discovered over 4000 exoplanets and candidates. However, current transit planet detection techniques based on the wavelet analysis and the Box Least Squares (BLS) algorithm have limited sensitivity in detecting minor planets with a low signal-to-noise ratio (SNR) and long periods with only 3-4 repeated signals over the mission lifetime of 4 years. This paper presents a novel precise-period transit signal detection methodology based on a new Graphics Processing Unit (GPU) Fast Folding algorithm in conjunction with a Convolutional Neural Network (CNN) to detect low SNR and/or long-period transit planet signals. A comparison with BLS is conducted on both simulated light curves and real data, demonstrating that the new method has higher speed, sensitivity, and reliability. For instance, the new system can detect transits with SNR as low as three while the performance of BLS drops off quickly around SNR of 7. Meanwhile, the GPU Fast Folding method folds light curves 25 times faster than BLS, a significant gain that allows exoplanet detection to occur at unprecedented period precision. This new method has been tested with all known transit signals with 100% confirmation. In addition, this new method has been successfully applied to the Kepler of Interest (KOI) data and identified a few new Earth-sized Ultra-short period (USP) exoplanet candidates and habitable planet candidates. The results highlight the promise for GPU Fast Folding as a replacement to the traditional BLS algorithm for finding small and/or long-period habitable and Earth-sized planet candidates in-transit data taken with Kepler and other space transit missions such as TESS(Transiting Exoplanet Survey Satellite) and PLATO(PLAnetary Transits and Oscillations of stars).

Keywords: algorithms, astronomy data analysis, deep learning, exoplanet detection methods, small planets, habitable planets, transit photometry

Procedia PDF Downloads 225

Authors: Anukriti Kumar, Tanmay Singh, Dinesh Kumar Vishwakarma

Abstract:

Traffic control has been one of the most common and irritating problems since the time automobiles have hit the roads. Problems like traffic congestion have led to a significant time burden around the world and one significant solution to these problems can be the proper implementation of the Intelligent Transport System (ITS). It involves the integration of various tools like smart sensors, artificial intelligence, position technologies and mobile data services to manage traffic flow, reduce congestion and enhance driver's ability to avoid accidents during adverse weather. Road and traffic signs’ recognition is an emerging field of research in ITS. Classification problem of traffic signs needs to be solved as it is a major step in our journey towards building semi-autonomous/autonomous driving systems. The purpose of this work focuses on implementing an approach to solve the problem of traffic sign classification by developing a Convolutional Neural Network (CNN) classifier using the GTSRB (German Traffic Sign Recognition Benchmark) dataset. Rather than using hand-crafted features, our model addresses the concern of exploding huge parameters and data method augmentations. Our model achieved an accuracy of around 97.6% which is comparable to various state-of-the-art architectures.

Keywords: multiclass classification, convolution neural network, OpenCV

Procedia PDF Downloads 176

Authors: Yufeng Li, Yinan Ma, Jing Wu, Chengnian Long

Abstract:

Infrared object detection aims at separating small and dim target from clutter background and its capabilities extend beyond the limits of visible light, making it invaluable in a wide range of applications such as improving safety, security, efficiency, and functionality. However, existing methods are usually sensitive to the noise of the input infrared image, leading to a decrease in target detection accuracy and an increase in the false alarm rate in high-noise environments. To address this issue, an infrared small target detection algorithm called YOLO-IR is proposed in this paper to improve the robustness to high infrared noise. To address the problem that high noise significantly reduces the clarity and reliability of target features in infrared images, we design a soft-threshold coordinate attention mechanism to improve the model’s ability to extract target features and its robustness to noise. Since the noise may overwhelm the local details of the target, resulting in the loss of small target features during depth down-sampling, we propose a deep and shallow feature fusion neck to improve the detection accuracy. In addition, because the generalized Intersection over Union (IoU)-based loss functions may be sensitive to noise and lead to unstable training in high-noise environments, we introduce a Wasserstein-distance based loss function to improve the training of the model. The experimental results show that YOLO-IR achieves a 5.0% improvement in recall and a 6.6% improvement in F1-score over existing state-of-art model.

Keywords: infrared small target detection, high noise, robustness, soft-threshold coordinate attention, feature fusion

Procedia PDF Downloads 73

Authors: Rakesh Kumar Mishra, Tarun Kumar Dan

Abstract:

Lead-Lag based Internal Model Control method is proposed based on Internal Model Control (IMC) strategy. In this paper, we have designed the Lead-Lag based Internal Model Control for binary distillation column for SISO process (considering only bottom product). The transfer function has been taken from Wood and Berry model. We have find the composition control and disturbance rejection using Lead-Lag based IMC and comparing with the response of simple Internal Model Controller.

Keywords: SISO, lead-lag, internal model control, wood and berry, distillation column

Procedia PDF Downloads 646

Authors: Anita Sadat Sadati Rostami, Reza Almasi Ghaleh

Abstract:

Surface electromyography (sEMG) offers important insights into muscle activation and has applications in fields including rehabilitation and human-computer interaction. The purpose of this work is to predict the degree of activation of two joints in the index finger using an LSTM-Transformer architecture trained on sEMG data from the Ninapro DB8 dataset. We apply advanced preprocessing techniques, such as multi-band filtering and customizable rectification methods, to enhance the encoding of sEMG data into features that are beneficial for regression tasks. The processed data is converted into spike patterns and simulated using Leaky Integrate-and-Fire (LIF) neuron models, allowing for neuromorphic-inspired processing. Our findings demonstrate that adjusting filtering parameters and neuron dynamics and employing the LSTM-Transformer model improves joint angle prediction performance. This study contributes to the ongoing development of deep learning frameworks for sEMG analysis, which could lead to improvements in motor control systems.

Keywords: surface electromyography, LSTM-transformer, spiking neural networks, hand kinematics, leaky integrate-and-fire neuron, band-pass filtering, muscle activity decoding

Procedia PDF Downloads 10

Authors: Vinicius G. Goecks, Gregory M. Gremillion, William D. Nothwang

Abstract:

Modern approaches to train intelligent agents rely on prolonged training sessions, high amounts of input data, and multiple interactions with the environment. This restricts the application of these learning algorithms in robotics and real-world applications, in which there is low tolerance to inadequate actions, interactions are expensive, and real-time processing and action are required. This paper addresses this issue introducing CyberSteer, a novel approach to efficiently design intrinsic reward functions based on human intention to guide deep reinforcement learning agents with no environment-dependent rewards. CyberSteer uses non-expert human operators for initial demonstration of a given task or desired behavior. The trajectories collected are used to train a behavior cloning deep neural network that asynchronously runs in the background and suggests actions to the deep reinforcement learning module. An intrinsic reward is computed based on the similarity between actions suggested and taken by the deep reinforcement learning algorithm commanding the agent. This intrinsic reward can also be reshaped through additional human demonstration or critique. This approach removes the need for environment-dependent or hand-engineered rewards while still being able to safely shape the behavior of autonomous robotic agents, in this case, based on human intention. CyberSteer is tested in a high-fidelity unmanned aerial vehicle simulation environment, the Microsoft AirSim. The simulated aerial robot performs collision avoidance through a clustered forest environment using forward-looking depth sensing and roll, pitch, and yaw references angle commands to the flight controller. This approach shows that the behavior of robotic systems can be shaped in a reduced amount of time when guided by a non-expert human, who is only aware of the high-level goals of the task. Decreasing the amount of training time required and increasing safety during training maneuvers will allow for faster deployment of intelligent robotic agents in dynamic real-world applications.

Keywords: human-robot interaction, intelligent robots, robot learning, semisupervised learning, unmanned aerial vehicles

Procedia PDF Downloads 259

Authors: H. Namazi, H. T. N. Kuan

Abstract:

It is known that in humans, the adaptation to a given odor occurs within a quite short span of time (typically one minute) after the odor is presented to the brain. Different models of human olfaction have been developed by scientists but none of these models consider the diffusion phenomenon in olfaction. A novel microscopic model of the human olfaction is presented in this paper. We develop this model by incorporating the transient diffusivity. In fact, the mathematical model is written based on diffusion of the odorant within the mucus layer. By the use of the model developed in this paper, it becomes possible to provide quantification of the objective strength of odor.

Keywords: diffusion, microscopic model, mucus layer, olfaction

Procedia PDF Downloads 505

Authors: Babak Bahri, Fatemeh Yassaee Meybodi, Changiz Eslahchi

Abstract:

In the pursuit of effective cancer therapies, the exploration of combinatorial drug regimens is crucial to leverage synergistic interactions between drugs, thereby improving treatment efficacy and overcoming drug resistance. However, identifying synergistic drug pairs poses challenges due to the vast combinatorial space and limitations of experimental approaches. This study introduces ClusterSyn, a machine learning (ML)-powered framework for classifying anti-cancer drug synergy scores. ClusterSyn employs a two-step approach involving drug clustering and synergy score prediction using a fully connected deep neural network. For each cell line in the training dataset, a drug graph is constructed, with nodes representing drugs and edge weights denoting synergy scores between drug pairs. Drugs are clustered using the Markov clustering (MCL) algorithm, and vectors representing the similarity of drug pairs to each cluster are input into the deep neural network for synergy score prediction (synergy or antagonism). Clustering results demonstrate effective grouping of drugs based on synergy scores, aligning similar synergy profiles. Subsequently, neural network predictions and synergy scores of the two drugs on others within their clusters are used to predict the synergy score of the considered drug pair. This approach facilitates comparative analysis with clustering and regression-based methods, revealing the superior performance of ClusterSyn over state-of-the-art methods like DeepSynergy and DeepDDS on diverse datasets such as Oniel and Almanac. The results highlight the remarkable potential of ClusterSyn as a versatile tool for predicting anti-cancer drug synergy scores.

Keywords: drug synergy, clustering, prediction, machine learning., deep learning

Procedia PDF Downloads 79

Authors: Linta Rose, Prasad K. Bhaskaran

Abstract:

Water-level information along the coast is very important for disaster management, navigation, planning shoreline management, coastal engineering and protection works, port and harbour activities, and for a better understanding of near-shore ocean dynamics. The water-level variation along a coast attributes from various factors like astronomical tides, meteorological and hydrological forcing. The study area is the Head Bay of Bengal which is highly vulnerable to flooding events caused by monsoons, cyclones and sea-level rise. The study aims to explore the extent to which wind and surface pressure can influence water-level elevation, in view of the low-lying topography of the coastal zones in the region. The ADCIRC hydrodynamic model has been customized for the Head Bay of Bengal, discretized using flexible finite elements and validated against tide gauge observations. Monthly mean climatological wind and mean sea level pressure fields of ERA Interim reanalysis data was used as input forcing to simulate water-level variation in the Head Bay of Bengal, in addition to tidal forcing. The output water-level was compared against that produced using tidal forcing alone, so as to quantify the contribution of meteorological forcing to water-level. The average contribution of meteorological fields to water-level in January is 5.5% at a deep-water location and 13.3% at a coastal location. During the month of July, when the monsoon winds are strongest in this region, this increases to 10.7% and 43.1% respectively at the deep-water and coastal locations. The model output was tested by varying the input conditions of the meteorological fields in an attempt to quantify the relative significance of wind speed and wind direction on water-level. Under uniform wind conditions, the results showed a higher contribution of meteorological fields for south-west winds than north-east winds, when the wind speed was higher. A comparison of the spectral characteristics of output water-level with that generated due to tidal forcing alone showed additional modes with seasonal and annual signatures. Moreover, non-linear monthly mode was found to be weaker than during tidal simulation, all of which point out that meteorological fields do not cause much effect on the water-level at periods less than a day and that it induces non-linear interactions between existing modes of oscillations. The study signifies the role of meteorological forcing under fair weather conditions and points out that a combination of multiple forcing fields including tides, wind, atmospheric pressure, waves, precipitation and river discharge is essential for efficient and effective forecast modelling, especially during extreme weather events.

Keywords: ADCIRC, head Bay of Bengal, mean sea level pressure, meteorological forcing, water-level, wind

Procedia PDF Downloads 221

Authors: Yanshuang Zhang, Byungho Jeong

Abstract:

There exists some time lag between the consumption of inputs and the production of outputs. This time lag effect should be considered in calculating efficiency of decision making units (DMU). Recently, a couple of DEA models were developed for considering time lag effect in efficiency evaluation of research activities. However, these models can’t discriminate efficient DMUs because of the nature of basic DEA model in which efficiency scores are limited to ‘1’. This problem can be resolved a super-efficiency model. However, a super efficiency model sometimes causes infeasibility problem. This paper suggests an output oriented super-efficiency model for efficiency evaluation under the consideration of time lag effect. A case example using a long term research project is given to compare the suggested model with the MpO model

Keywords: DEA, Super-efficiency, Time Lag, research activities

Procedia PDF Downloads 658

Authors: Amdy Moustapha Drame, Ilya V. Germashev, E. A. Markushevskaya

Abstract:

Among the main problems of medicine is breast cancer, from which a significant number of women around the world are constantly dying. Therefore, the detection of malignant breast tumors is an urgent task. For many years, various technologies for detecting these tumors have been used, in particular, in thermal imaging in order to determine different levels of breast cancer development. These periodic screening methods are a diagnostic tool for women and may have become an alternative to older methods such as mammography. This article proposes a model for the identification of malignant neoplasms of the mammary glands by the asymmetry of internal and external thermal imaging fields.

Keywords: asymmetry, breast cancer, tumors, deep learning, thermogram, convolutional transformation, classification

Procedia PDF Downloads 60

Authors: Zainab Magaji Musa, Nordin M. A. Rahman, Julaily Aida Jusoh

Abstract:

The web services applications for digital reference service (WSDRS) of LIS model is an informal model that claims to reduce the problems of digital reference services in libraries. It uses web services technology to provide efficient way of satisfying users’ needs in the reference section of libraries. The formal WSDRS model consists of the Z specifications of all the informal specifications of the model. This paper discusses the formal validation of the Z specifications of WSDRS model. The authors formally verify and thus validate the properties of the model using Z/EVES theorem prover.

Keywords: validation, verification, formal, theorem prover

Procedia PDF Downloads 516

Authors: Vahid Bayrami Rad

Abstract:

In recent years, the use of artificial intelligence techniques, tools, and methods in processing medical images and health-related applications has been highlighted and a lot of research has been done in this regard. For example, colonoscopy and diagnosis of colon lesions are some cases in which the process of diagnosis of lesions can be improved by using image processing and artificial intelligence algorithms, which help doctors a lot. Due to the lack of accurate measurements and the variety of injuries in colonoscopy images, the process of diagnosing the type of lesions is a little difficult even for expert doctors. Therefore, by using different software and image processing, doctors can be helped to increase the accuracy of their observations and ultimately improve their diagnosis. Also, by using automatic methods, the process of diagnosing the type of disease can be improved. Therefore, in this paper, a deep learning framework called KCBA is proposed to classify colonoscopy lesions which are composed of several methods such as K-means clustering, a bag of features and deep auto-encoder. Finally, according to the experimental results, the proposed method's performance in classifying colonoscopy images is depicted considering the accuracy criterion.

Keywords: colorectal cancer, colonoscopy, region of interest, narrow band imaging, texture analysis, bag of feature

Procedia PDF Downloads 57

Authors: George Zhou, Yunchan Chen, Candace Chien

Abstract:

Kidney replacement therapy is the current standard of care for end-stage renal diseases. In-center or home hemodialysis remains an integral component of the therapeutic regimen. Arteriovenous fistulas (AVF) make up the vascular circuit through which blood is filtered and returned. Naturally, AVF patency determines whether adequate clearance and filtration can be achieved and directly influences clinical outcomes. Our aim was to build a deep learning model for automated AVF stenosis screening based on the sound of blood flow through the AVF. A total of 311 patients with AVF were enrolled in this study. Blood flow sounds were collected using a digital stethoscope. For each patient, blood flow sounds were collected at 6 different locations along the patient’s AVF. The 6 locations are artery, anastomosis, distal vein, middle vein, proximal vein, and venous arch. A total of 1866 sounds were collected. The blood flow sounds are labeled as “patent” (normal) or “stenotic” (abnormal). The labels are validated from concurrent ultrasound. Our dataset included 1527 “patent” and 339 “stenotic” sounds. We show that blood flow sounds vary significantly along the AVF. For example, the blood flow sound is loudest at the anastomosis site and softest at the cephalic arch. Contextualizing the sound with location metadata significantly improves classification performance. How to encode and incorporate categorical metadata is an active area of research1. Herein, we study ordinal (i.e., integer) encoding schemes. The numerical representation is concatenated to the flattened feature vector. We train a vision transformer (ViT) on spectrogram image representations of the sound and demonstrate that using scalar multiples of our integer encodings improves classification performance. Models are evaluated using a 10-fold cross-validation procedure. The baseline performance of our ViT without any location metadata achieves an AuROC and AuPRC of 0.68 ± 0.05 and 0.28 ± 0.09, respectively. Using the following encodings of Artery:0; Arch: 1; Proximal: 2; Middle: 3; Distal 4: Anastomosis: 5, the ViT achieves an AuROC and AuPRC of 0.69 ± 0.06 and 0.30 ± 0.10, respectively. Using the following encodings of Artery:0; Arch: 10; Proximal: 20; Middle: 30; Distal 40: Anastomosis: 50, the ViT achieves an AuROC and AuPRC of 0.74 ± 0.06 and 0.38 ± 0.10, respectively. Using the following encodings of Artery:0; Arch: 100; Proximal: 200; Middle: 300; Distal 400: Anastomosis: 500, the ViT achieves an AuROC and AuPRC of 0.78 ± 0.06 and 0.43 ± 0.11. respectively. Interestingly, we see that using increasing scalar multiples of our integer encoding scheme (i.e., encoding “venous arch” as 1,10,100) results in progressively improved performance. In theory, the integer values do not matter since we are optimizing the same loss function; the model can learn to increase or decrease the weights associated with location encodings and converge on the same solution. However, in the setting of limited data and computation resources, increasing the importance at initialization either leads to faster convergence or helps the model escape a local minimum.

Keywords: arteriovenous fistula, blood flow sounds, metadata encoding, deep learning

Procedia PDF Downloads 88

Authors: Shayan Mohajer Hamidi

Abstract:

Federated learning has emerged as a promising approach for distributed model training while preserving data privacy. However, the challenges of communication overhead, limited network resources, and slow convergence hinder its widespread adoption. On the other hand, knowledge distillation has shown great potential in compressing large models into smaller ones without significant loss in performance. In this paper, we propose an innovative framework that combines federated learning and knowledge distillation to address these challenges and enhance the efficiency of distributed learning. Our approach, called Federated Knowledge Distillation (FKD), enables multiple clients in a federated learning setting to collaboratively distill knowledge from a teacher model. By leveraging the collaborative nature of federated learning, FKD aims to improve model compression while maintaining privacy. The proposed framework utilizes a coded teacher model that acts as a reference for distilling knowledge to the client models. To demonstrate the effectiveness of FKD, we conduct extensive experiments on various datasets and models. We compare FKD with baseline federated learning methods and standalone knowledge distillation techniques. The results show that FKD achieves superior model compression, faster convergence, and improved performance compared to traditional federated learning approaches. Furthermore, FKD effectively preserves privacy by ensuring that sensitive data remains on the client devices and only distilled knowledge is shared during the training process. In our experiments, we explore different knowledge transfer methods within the FKD framework, including Fine-Tuning (FT), FitNet, Correlation Congruence (CC), Similarity-Preserving (SP), and Relational Knowledge Distillation (RKD). We analyze the impact of these methods on model compression and convergence speed, shedding light on the trade-offs between size reduction and performance. Moreover, we address the challenges of communication efficiency and network resource utilization in federated learning by leveraging the knowledge distillation process. FKD reduces the amount of data transmitted across the network, minimizing communication overhead and improving resource utilization. This makes FKD particularly suitable for resource-constrained environments such as edge computing and IoT devices. The proposed FKD framework opens up new avenues for collaborative and privacy-preserving distributed learning. By combining the strengths of federated learning and knowledge distillation, it offers an efficient solution for model compression and convergence speed enhancement. Future research can explore further extensions and optimizations of FKD, as well as its applications in domains such as healthcare, finance, and smart cities, where privacy and distributed learning are of paramount importance.

Keywords: federated learning, knowledge distillation, knowledge transfer, deep learning

Procedia PDF Downloads 75

Authors: SKhan Park, Seok Jin Lee, Jong Sun Kim, Jun Ho Lee, Bong Chan Kim

Abstract:

Because of the large economic losses caused by traffic congestion in metropolitan areas, various studies on the underground network design and construction techniques has been performed various studies in the developed countries. In Korea, it has performed a study to develop a versatile double-deck of deep tunnel model. This paper is an introduction to develop a ground reinforcement method to enable the safe tunnel construction in the weakened pillar section like as junction of tunnel. Applicable ground reinforcement method in the weakened section is proposed and it is expected to verify the method by the field application tests.

Keywords: double-deck tunnel, ground reinforcement, tunnel construction, weakened pillar section

Procedia PDF Downloads 409

Authors: Radzuan Junin, Dayang Zulaika A. Hasbollah

Abstract:

The increasing amount of greenhouse gasses in the atmosphere recently has become one of the discussed topics in relation with world’s concern on climate change. Developing countries’ emissions (such as Malaysia) are now seen to surpass developed country’s emissions due to rapid economic development growth in recent decades. This paper presents the potential storage sites suitability and storage capacity assessment for CO2 sequestration in sedimentary basins of Malaysia. This study is the first of its kind that made an identification of potential storage sites and assessment of CO2 storage capacity within the deep saline aquifers in the country. The CO2 storage capacity in saline formation assessment was conducted based on the method for quick assessment of CO2 storage capacity in closed, and semi-closed saline formations modified to suit the geology setting of Malaysia. Then, an integrated approach that involved geographic information systems (GIS) analysis and field data assessment was adopted to provide the potential storage sites and its capacity for CO2 sequestration. This study concentrated on the assessment of major sedimentary basins in Malaysia both onshore and offshore where potential geological formations which CO2 could be stored exist below 800 meters and where suitable sealing formations are present. Based on regional study and amount of data available, there are 14 sedimentary basins all around Malaysia that has been identified as potential CO2 storage. Meanwhile, from the screening and ranking exercises, it is obvious that Malay Basin, Central Luconia Province, West Baram Delta and Balingian Province are respectively ranked as the top four in the ranking system for CO2 storage. 27% of sedimentary basins in Malaysia were evaluated as high potential area for CO2 storage. This study should provide a basis for further work to reduce the uncertainty in these estimates and also provide support to policy makers on future planning of carbon capture and sequestration (CCS) projects in Malaysia.

Keywords: CO₂ storage, deep saline aquifer, GIS, sedimentary basin

Procedia PDF Downloads 360

Authors: Sapan Bhatnagar, Danish Akhtar, Salman Ahmed, Asif Ekbal, Gufran Beig

Abstract:

Particulates are the solid and liquid droplets present in the atmosphere, they have serious negative effects on human health and environment. PM10 and PM2.5 are so small that they can penetrate deep into our lungs through the respiratory system. Determination of the amount of particulates present in the atmosphere per cubic meter is necessary to monitor, regulate and model atmospheric particulate levels. Air Quality Index is an index tells us how clean or polluted our air is, and what associated health effects might be a concern for us. The AQI focuses on health affects you may experience within a few hours or days after breathing polluted air. The quality rating for each pollutant was calculated. The geometric mean of these quality ratings gives the Air Quality Index. The existing concentrations of pollutants were compared with ambient air quality standards.

Keywords: air quality index, particulate, respirable dust sampler, dust sampler

Procedia PDF Downloads 576

Authors: Federico Garrido, Mostafa El Hayani, Ahmed Shams

Abstract:

The exploration of the potential of artificial intelligence (AI) in architecture is still embryonic, however, its latent capacity to change design disciplines is significant. 'Synthetic Classism' is a research project that questions the underlying aspects of classically organized architecture not just in aesthetic terms but also from a geometrical and morphological point of view, intending to generate new architectural information using historical examples as source material. The main aim of this paper is to explore the uses of artificial intelligence and machine learning algorithms in architectural design while creating a coherent narrative to be contained within a design process. The purpose is twofold: on one hand, to develop and train machine learning algorithms to produce architectural information of small pavilions and on the other, to synthesize new information from previous architectural drawings. These algorithms intend to 'interpret' graphical information from each pavilion and then generate new information from it. The procedure, once these algorithms are trained, is the following: parting from a line profile, a synthetic 'front view' of a pavilion is generated, then using it as a source material, an isometric view is created from it, and finally, a top view is produced. Thanks to GAN algorithms, it is also possible to generate Front and Isometric views without any graphical input as well. The final intention of the research is to produce isometric views out of historical information, such as the pavilions from Sebastiano Serlio, James Gibbs, or John Soane. The idea is to create and interpret new information not just in terms of historical reconstruction but also to explore AI as a novel tool in the narrative of a creative design process. This research also challenges the idea of the role of algorithmic design associated with efficiency or fitness while embracing the possibility of a creative collaboration between artificial intelligence and a human designer. Hence the double feature of this research, both analytical and creative, first by synthesizing images based on a given dataset and then by generating new architectural information from historical references. We find that the possibility of creatively understand and manipulate historic (and synthetic) information will be a key feature in future innovative design processes. Finally, the main question that we propose is whether an AI could be used not just to create an original and innovative group of simple buildings but also to explore the possibility of fostering a novel architectural sensibility grounded on the specificities on the architectural dataset, either historic, human-made or synthetic.

Keywords: architecture, central pavilions, classicism, machine learning

Procedia PDF Downloads 140

Authors: Matthew C. Best

Abstract:

Linear Multi-Input Multi-Output (MIMO) dynamic models can be identified, with no a priori knowledge of model structure or order, using a new Generalised Identifying Filter (GIF). Based on an Extended Kalman Filter, the new filter identifies the model iteratively, in a continuous modal canonical form, using only input and output time histories. The filter’s self-propagating state error covariance matrix allows easy determination of convergence and conditioning, and by progressively increasing model order, the best fitting reduced-order model can be identified. The method is shown to be resistant to noise and can easily be extended to identification of smoothly nonlinear systems.

Keywords: system identification, Kalman filter, linear model, MIMO, model order reduction

Procedia PDF Downloads 594

Authors: Ali Hamad, Ibrahim Al-Drous, Saleh Al-Jufout

Abstract:

This paper presents a case study of using STATCOM to enhance the performance of Al-Qatraneh 33-kV transmission line. The location of the STATCOM was identified maintaining minimum voltage drops at the 110 load nodes. The transmission line and the 110 load nodes have been modeled by MATLAB/Simulink. The suggested STATCOM and its location will increase the transmission capability of this transmission line and overcome the overload expected in the year 2020. The annual percentage loading rise has been considered as 14%. A graphical representation of the line voltages and the voltage drops at different load nodes has been illustrated.

Keywords: FACTS, MATLAB, STATCOM, transmission line, voltage drop

Procedia PDF Downloads 443

Authors: Mounir Zekkaoui, Abdelhadi Fennan

Abstract:

The software evolution control requires a deep understanding of the changes and their impact on different system heterogeneous artifacts. And an understanding of descriptive knowledge of the developed software artifacts is a prerequisite condition for the success of the evolutionary process. The implementation of an evolutionary process is to make changes more or less important to many heterogeneous software artifacts such as source code, analysis and design models, unit testing, XML deployment descriptors, user guides, and others. These changes can be a source of degradation in functional, qualitative or behavioral terms of modified software. Hence the need for a unified approach for extraction and representation of different heterogeneous artifacts in order to ensure a unified and detailed description of heterogeneous software artifacts, exploitable by several software tools and allowing to responsible for the evolution of carry out the reasoning change concerned.

Keywords: heterogeneous software artifacts, software evolution control, unified approach, meta model, software architecture

Procedia PDF Downloads 446