University of Technology Malaysia
Universiti Teknologi Malaysia Institutional RepositoryNot a member yet
70456 research outputs found
Sort by
Development of mobile application for detection and grading of diabetic retinopathy
No full textThe key to preventing blindness caused by diabetic retinopathy (DR) is regular screening and early recognition during its early stages. Currently, DR grading is done manually by ophthalmologists and trained graders where the process is time-consuming. Therefore, this paper aims to develop a mobile app that can provide DR detection and grading without a professional or doctor. The patients will be referred to ophthalmologists if further evaluations are required. This research builds an image classification within a mobile application by using deep learning techniques which utilized the Google AI technologies: Google TensorFlow and Google Cloud Platform (Cloud AutoML and Cloud storage). Image classification is performed in two layers which involve DR detection and grading. A total of 12,062 fundus images are chosen from the dataset collected and undergo image preprocessing. The preprocessed images are used to train the model in TensorFlow and Cloud AutoML, respectively. The model will be implemented into the mobile application after being trained with high accuracy. The final test accuracy for the MobileNet pretrained model is 82.9%, while averaging precision for the model of Cloud AutoML is 75%. Further research is required to improve the stability of this algorithm and mobile app for real clinical environment settings
Characterization of the first mitogenomes of the smallest fish in the world, paedocypris progenetica, from peat swamp of Peninsular Malaysia, Selangor, and Perak
Get PDFThe two complete mitochondrial genomes (mitogenomes) of Paedocypris progenetica, the smallest fish in the world which belonged to the Cyprinidae family, were sequenced and assembled. The circular DNA molecules of mitogenomes P1-P. progenetica and S3-P. pro-genetica were 16,827 and 16,616 bp in length, respectively, and encoded 13 protein-cod-ing genes, 22 transfer RNA genes, two ribosomal RNA genes, and one control region. The gene arrangements of P. progenetica were identical to those of other Paedocypris species. BLAST and phylogenetic analyses revealed variations in the mitogenome sequences of two Paedocypris species from Perak and Selangor. The circular DNA molecule of P. progenetica yield a standard vertebrate gene arrangement and an overall nucleotide composition of A 33.0%, T 27.2%, C 23.5%, and G 15.5%. The overall AT content of this species was consis-tent with that of other species in other genera. The negative GC-skew and positive AT-skew of the control region in P. progenetica indicated rich genetic variability and AT nucleotide bias, respectively. The results of this study provide genomic variation information and enhance the understanding of the mitogenome of P. progenetica. They could later deliver highly valuable new insight into data for phylogenetic analysis and population genetics
Perfect codes in the spanning and induced subgraphs of the unity product graph
Get PDFThe unity product graph of a ring R is a graph which is obtained by setting the set of unit elements of R as the vertex set. The two distinct vertices ri and rj are joined by an edge if and only if ri · rj = e. The subgraphs of a unity product graph which are obtained by the vertex and edge deletions are said to be its induced and spanning subgraphs, respectively. A subset C of the vertex set of induced (spanning) subgraph of a unity product graph is called perfect code if the closed neighbourhood of c, S1 (c) forms a partition of the vertex set as c runs through C. In this paper, we determine the perfect codes in the induced and spanning subgraphs of the unity product graphs associated with some commutative rings R with identity. As a result, we characterize the rings R in such a way that the spanning subgraphs admit a perfect code of order cardinality of the vertex set. In addition, we establish some sharp lower and upper bounds for the order of C to be a perfect code admitted by the induced and spanning subgraphs of the unity product graphs
Photodegradation of reactive blue 4 using suspension of anatase titanium dioxide and corn cob
No full textTextile dyeing often employs reactive dyes. The dye wastewater contains hazardous materials and is toxic to humans and the environment. Photodegradation using a semiconductor photocatalyst is a promising alternative approach for water purification and wastewater treatment. However, the photocatalyst’s low adsorption ability is a problem in the photocatalysis process. To compensate for this shortcoming, photocatalyst content must be combined with an adsorbent. Raw corn cob and titanium dioxide (TiO2) were used in this photocatalysis. Due to a synergistic impact, raw corn cob’s ability to adsorb and titanium dioxide’s ability to photodegrade organic pollutants from water bodies is expected to boost the removal performance. The degradation of Reactive Blue 4 (RB4) as a targeted dye was carried out in this research using a suspended mixture of commercial anatase TiO2 and raw corn cob under UV light. The effect of initial pH solution, initial dye concentration and contact time, TiO2-corn cob dosages, and the influence of other pollutants were investigated as factors influencing photodegradation-adsorption of RB4. FTIR and SEM analyses were performed to characterize the prepared materials. The high removal rate of RB4 was obtained at a low pH of 2 and RB4 concentration of 40 ppm. The increased dose of TiO2-corn cob improved the RB4 dye removal performance. The optimum percentage removal of RB4 was 92.65 % at pH 2, 40 ppm of RB4 concentration mixed with 1.2 g of TiO2-corn cob in 60 minutes of UV light irradiation. SEM observation revealed that corn cob surfaces are uneven and very porous in nature. FTIR test indicate the presence of functional group on the TiO2-corn cob helps in the adsorption of the RB4. In conclusion, combining photodegradation and adsorption systems as a hybrid treatment method resulted in a synergistic increase in RB4 removal performance
An FPGA-based IP core subscription-oriented fog computing platform
No full textThis paper presents an FPGA-based IP core subscription-oriented fog platform. The application plane of the proposed platform can be dynamically reconfigured and defined remotely. Besides, the proposed platform includes a PUF-based IP core licensing and metering mechanisms to enable application IP core subscription. Thus, the application IP core developer has total controllability over the application plane on the proposed platform as the PUF circuitry is placed within the reconfigurable partition. The proposed platform is implemented and tested experimentally on NetFPGA CML with a time-series anomaly detection analytics as case studies. The proposed platform with hosted analytics utilized < 43 % logic resources in a Kintex 7 FPGA device (XC7K325T-IFFG676). The throughput and latency of implemented FPGA-based time-series anomaly detection analytics are 3096 detections per second and 0.32 ms, respectively, which is more than 61 × speedup over the reference analytics implemented in software. The customized DPR controller is implemented internally within the FPGA device with a reconfiguration time of 15.34 ms
Compatible linear lypunov function for infinite dimensional volterra quadratic stochastic operators
No full textThe simplest non-linear operator is the quadratic ones. Most of the researches in this direction were investigating on finite set of all probability distributions. However, there are models where the probability distributions are countably infinite, which means that the considered operators are defined on infinite-dimensional spaces. We restrict ourselves to Quadratic Stochastic Operators (QSOs) define on infinite dimension, specifically a class of QSOs called Volterra. In this paper, we construct a linear Lyapunov function for infinite dimensional Volterra QSOs by means of finite dimensional ones
Assessing clinical usefulness of readmission risk prediction model
No full textReadmission manifests signs of degraded quality of care and increased healthcare cost. Such adverse event may be attributed to premature discharge, unsuccessful treatments, or worsening comorbidities. Predictive modeling provides useful information to identify patients at a higher risk for readmission for targeted interventions. Though many studies have proposed readmission risk predictive models and validated their discriminative ability with performance metrics, few examined the net benefit realized by a predictive model. We compared traditional logistic regression against modern neural network to predict unplanned readmission. An added value of 7% on discriminative ability is observed for modern machine learning model compared to regression. A cost analysis is provided to assist physicians and hospital management for translating the theoretical value into real cost and resource allocation after model implementation. The neural network model is projected to contribute 15× more savings by reducing readmissions. Aside from constructing better performing models, the results of our study demonstrate the potential of a clinically helpful prediction tool in terms of strategies to reduce cost associated with readmission
3D tomogram using on-chip ECT with ac-based capacitance measuring system
No full textThe emergence of microfluidics technology has inspired the development of on-chip tomogram to replace the bulky microscopic methods. This paper proposes an on-chip 3D imaging technique using an electrical capacitance tomography (ECT) with an AC-based measuring system. To accomplish the sensing of the 34 electrodes equipped by the on-chip ECT, an AC-based measuring system that comprises 34 measuring channels with the measuring frequency of 10 kHz was developed. The system was designed using the commercially available analogue integrated circuits (ICs), which suits well for the common laboratorial usage with low fabrication cost. To validate the AC-based system, its performance was assessed based on the accuracy, signal-to-noise ratio (SNR), and standard deviation, using the standard capacitors ranging from 1 pF to 10 nF. The assessment showed that the accuracy was high with relative error around 6% for capacitance values higher than 10 pF. It was found that the minimum SNR was 12.98 dB and the maximum standard deviation was 7.37 pF. Subsequently, employing the developed AC-based system, the 3D images of a cubic agar with encapsulated Saccharomyces cerevisiae cells, were reconstructed successfully by the on-chip ECT. The evaluation results suggest that the presented design offers a potential solution for the on-chip 3D imaging
Lightweight IoT based indoor positioning for guard touring system
No full textAn Indoor Positioning System (IPS) with Internet of Things (IoT) platform for a Guard Touring System (GTS) application is developed to track in real-time the guards’ whereabouts in an indoor setting when they perform their patrolling duties. The developed system comprises of Bluetooth Low Energy (BLE) beacons, mobile application and Node-RED IoT platform. BLE beacons are used to collect position data. The position information from the beacons captured by the developed mobile phone application is then transmitted using MQTT broker service to reach the Node-RED cloud platform for analysis of the information and generating a real-time end-user display dashboard. The real-time position data is also stored in MongoDB database platform for future reference. Two methods are used to estimate the indoor positioning of the guards which are machine learning using Linear Regression model and BLE Media Access Control (MAC) Identifier. The findings show the BLE MAC Identifier method provides a high accuracy of 98% and the least delay in decision time, which can be as fast as 0.5 s. The method is also more cost-effective as it uses lesser number of devices to achieve high accuracy indoor positioning estimation
User models of students social interaction in e-learning based on learning orientation profile
No full textResearch on user modelling has blossomed in recent years. The user model is designed to represent characteristics of users or students, including preferences, knowledge, competencies, tasks and objectives. The user model helps instructors to monitor students’ learning processes and to clearly see the outcome and effectiveness of the learning. Therefore, it is necessary to fill in the gap of no research done on user modelling for each Learning Orientations Profile with regard to social interactions in e-learning. Therefore, this research was done with the objectives to investigate students’ social interactions in e-learning and synthesize a user model for each learner’s profile based on students’ social interactions in e-learning. Accordingly, this research was done by referring quantitative research design that follows inferential approach. Therefore, firstly, students are categorized into four Learning Orientations Profiles, which are Transformance, Performance, Conformance and Resistance. All students learned through the same e-learning system for fifteen weeks. Then, the number of frequencies of their social interactions in e-learning are analyzed through the e-learning database. The interactions are including students’ involvement in social interactivity functions such as forum, chatting, and messaging. The data analyzed is then synthesized into a user model for each Learning Orientations Profile. Thus, four user models for all four Learning Orientations Profiles are the expected output of this research. Significantly, in order to emphasize individual difference in learning, the user models could be used by instructors or researchers to design and develop social interactivity functions that suitable for each Learning Orientations Profile