Search

COVID-19 pandemic has affected everyone in terms of the life style, the way of carrying out work, meetings, shopping, teaching and learning and many more. Many economy sectors were also affected. This paper is reporting on how the institute of higher learning transforming the usual physical open day events to the virtual environment due to the Movement Control Order (MCO) imposed by the Malaysian Government in the fight to contain the spreading of the COVID-19 virus among the community. The virtual open day platform was created entirely by 3D visualized rendering software. It features plenty of stunning and conceptual virtual spaces such as the Grand Lobby and up to 16 program booths. The platform is also integrated with interactive and immersive features for viewers to navigate, explore and click on any hotspot. Clicking on a hotspot triggers an action such as opening a photo, poster, video, text file, or URL link. Google Analytics is being integrated to analyze the statistics and demographic of the viewers/visitors. The virtual platform is highly customizable and can be used to host other commercialized events such as virtual exhibitions, virtual fair, virtual shopping experience, and virtual travel tours. This is why virtual tours are considered the future of creativity with control in the palm of our hands.

In this paper, a deep learning technique is used to segment skin lesions of Atopic Eczema. A five-layer model based on U-Net architecture is trained to perform segmentation on a dataset with 84 images of Atopic Eczema. The image dimension is set to 512 by 512 pixels and grayscale, red, green and blue color channels are used for segmentation. Two parameters, accuracy and loss serve as the monitoring parameters during the training of the model. For performance evaluation, parameters including accuracy, sensitivity, specificity, Positive Predictive Value (PPV) and Negative Predictive Value (NPV) are used. An average validation accuracy of 85.16% is obtained by using green channel.

To keep up with the dominance in the field of leading semiconductor technology innovation, TSMC has announced the risk production of its most advanced 5nm CMOS logic node [1] using the full-fledged EUV and high mobility channel (HMC) FinFETs. Supporting the state of the art mobile SOC chips and HPC application needs, this 5nm technology node provides ~1.8x improvement in logic density, 15% speed gain and 30% power reduction as compared to its previous 7nm generation [1] - [2] . This paper, for the first time, brings out the detailed reliability attributes for the TSMC 5nm technology node from the device to various chip/package level reliability testing. While delivering an enhanced performance for N5, here we demonstrate the reliability benefit in terms of Bias Temperature Instability (BTI), Hot Carrier Degradation (HCD), Time dependent dielectric breakdown (TDDB) degradation modes in device level. Not only that, N5 node also delivers excellent reliability margins in chip level evaluations for SRAM and logic CPU/GPU applications. Moreover, to compete with the strictest industry reliability requirements even to the DPPM or low DPPB levels, N5 logic confirms its benefit for the automotive applications.

In this paper our aim is to develop a fully automated eczema skin lesion segmentation method. We have studied three supervised learning methods for segmentation of lesions: Support Vector Machine (SVM), Naive Bayesian Classifier (NBC) and K-Nearest Neighbor (KNN). Two sets of images that are different in erythema severity levels (mild, moderate) are used for training the supervised classifiers. From the training images 108 features are extracted that are ranked using four feature ranking methods (standard deviation, T-statistical score, fisher scoring, and correlation coefficient) to obtain the most significant features. The performance of classification is investigated using green channel of RGB and CSN-I RGB color space. The performance of the different methods is assessed by comparing the segmented lesions with the gold standard segmented images. Based on these comparisons, it is observed that SVM classifier shows the best segmentation result having an accuracy of 84.43% whereas the accuracy of NBC and KNN is 82.77% and 83.53% respectively.

Identifying anomalous samples from highly complex and unstructured data is a crucial but challenging task in a variety of intelligent systems. In this paper, we present a novel deep anomaly detection framework named AnoDM (standing for Anomaly detection based on unsupervised Disentangled representation learning and Manifold learning). The disentanglement learning is currently implemented by β-VAE for automatically discovering interpretable factorized latent representations in a completely unsupervised manner. The manifold learning is realized by t-SNE for projecting the latent representations to a 2D map. We define a new anomaly score function by combining β-VAE’s reconstruction error in the raw feature space and local density estimation in the t-SNE space. AnoDM was evaluated on both image and time-series data and achieved better results than models that use just one of the two measures and other deep learning methods.

The combination of cognitive mm-Wave Dual-Beam (DB) MIMO Phased Array systems with optical sensing solutions is proposed, targeting new functionalities for environmental perception and ubiquitous interactions. The resulting paradigms exploit the fusion of mm-Wave and optical sensing solutions to enable emerging technologies facilitating interactions of humans with smart devices and systems in randomly changing environments. Perspectives for autonomous vehicles with Advanced Driver Assistance Systems (ADAS), including Gesture-Recognition (GR), through ubiquitous interactions based on hybrid cognitive mm-Wave RFIC technologies and optical systems, are drawn. Several hardware realizations of mm-Wave phased-arrays are built and co-assembled with optical systems for smart data fusion and real-time signal processing for an autonomous secure decision-making process. A unified modeling and measurement platform is proposed with the concept of Multi-Physics (EM-thermal-mechanical) Numerical-Co-Simulation-Clone (NCSC), seen as the counterpart of the functional hardware, which enables 4D (space-time evolution) of augmented reality.

Despite recent advances, low-voltage operation remains one of the key approaches for power reduction. However, the continuous scaling of the SRAM bit cell, in advanced technologies, based on the transistor's minimum geometry is accompanied by increased random threshold voltage variations that limit the SRAM's minimum operating voltage (V MIN ). The introduction of FinFET transistors has provided better short channel effects and less random dopant fluctuation, compared to prior bulk CMOS technology. However, the quantization of transistor sizing continues to be a major challenge for high-density 6T SRAM design. Figure 15.1.1(a) shows the layout of a high-density 6T SRAM bit cell in a 5nm EUV and high-mobility channel FinFET technology. In order to achieve a compact bit-cell area, using FinFET technology, all transistors have to consist of a single fin. Figure 15.1.1(b) shows the SRAM bit-cell area scaling trend since 2011: area continues to scale, even as technology evolved into the sub-20nm era. This work reports a 0.021 µm2 high-density SRAM bit cell for a leading edge 5nm technology [1]. Figure 15.1.2(a) shows the SRAM voltage scaling trend: where the blue line indicates the V MIN of the SRAM bit cell without circuit assist and the red line indicates the V MIN with circuit assist. Write assist is the key enabler for lowering V DD operation in 5nm technology. Figure 15.1.2(b) shows a contention between pull-up (PU) and pass-gate (PG) transistors during a write operation. A stronger PU transistor yields higher read stability, but it degrades the write margin significantly and results in a contention write V MIN issue. In order to improve the write V MIN , the negative BL (NBL) and lower cell V DD (LCV) techniques have been proposed [2], [6]. However, the write-assist circuitry results in area overhead. In order to reduce the area of the assist circuitry, this work proposes NBL [7] and LCV with metal coupling/charge sharing capacitors to reduce the area overhead of the NBL-boosted and LCV charge-sharing capacitors. Furthermore, the flying BL (FBL) architecture [8] is also implemented to further reduce area.

A leading edge 5nm CMOS platform technology has been defined and optimized for mobile and HPC applications. This industry-leading 5nm technology features, for the first time, full-fledged EUV, and high mobility channel (HMC) finFETs with densest 0.021µm2 HD SRAM. This true 5nm CMOS platform technology is a full node scaling from our successful 7nm node [4] in offering ~1.84x logic density, 15% speed gain or 30% power reduction. The 5nm platform technology successfully passed qualification [3] with consistently high yielding 256Mb HD/HC SRAM, and large logic test chip consisting of CPU/GPU/SoC blocks. Currently in risk production, this true 5nm platform technology is on schedule for high volume production in 1H 2020.

The LabVIEW-based educational atomic force microscope (AFM) with National Instruments Educational Laboratory Virtual Instrumentation Suite (NI ELVIS) as the controller platform was built previously. However, it has limited function. Modification is needed to enhance the graphical user interface (GUI) so that more advanced features can be integrated into the existing platform. In this study, the force measurement function is introduced into the existing AFM system. Other additional features such as calibration of the cantilever spring constant and data storage are integrated into the system as well.

Large scale adoption of electric vehicles (EV) has yet to be accelerated due to fundamental challenges such as the availability of charging infrastructure, mileage of the EV, as well as the higher lifetime costs associated with the batteries. In order to address these challenges, this paper introduces a framework to model, analyze and visualize the driving-related energy performance of electric vehicles. Based on a case study of real-world data inputs of throttle, brake and road conditions in Singapore, the proposed framework illustrates several findings in terms of modelling accuracy and driving behaviour identification together with visualization and simulation results in open-source Simulation of Urban Mobility software.

A robust battery management system (BMS) is important for electric vehicle (EV) to preserve the battery health. As EV is gaining popularity as a mode of public transport, the fleet operator should be concerned of optimizing the usage of a fully charged battery, this boils down to the driving behavior. This paper investigates the relation of driving parameters to the battery discharge rate. The driving behavior is associated with vehicle speed, motor speed, throttle position and brake pressure. The data is collected from experiment where the EV is operated through urban roads in Singapore. Principal Component Analysis (PCA) is used to reduce dimensionality of data. We identified an interesting feature in the data distribution, which can be quantified by centroid. Using this feature, we conduct the regression analysis and find out that the centroid has a linear relationship with the duration of battery retaining its state of charge (SOC), which is supported by Pearson correlation coefficient of 0.89 obtained in the analysis.

In this paper, an enhanced face recognition method is proposed to develop an efficient student attendance system. In the face recognition stage, enhanced local binary pattern (ELBP) is implemented by increasing the radius of the original local binary pattern (LBP) operator and principal component analysis (PCA) is used to extract the features from facial images. The ELBP outperforms the original LBP by suppressing the adverse influence of different illumination conditions and thus results in increased recognition rate. The test recognition rate on the Yale face database of the original LBP is 81.9%, while the ELBP with radius 2 to 5 pixels has achieved face recognition rate of 84.6% to 88.5%. The face images are recognized based on the classification results obtained from the combination of algorithms (PCA and ELBP). A user-friendly interface has been designed to improve the attendance marking process. Image pre-processing is performed to ensure better performance of the system. Lastly, the attendance of the recognized students is marked accordingly and saved in an excel sheet. Students who have not undergone pre-registration can register on the spot. Notification will be generated if a student signs in more than once. The recognition rate for high quality images in the attendance system is 100% whereas for low quality images it is 92.31%.

Herewith the efficiency of various configurations of employing Kalman filter algorithm for on-the-fly pre-processing of the sensory network originated data streams in the Internet of Things (IoT) systems is investigated. Contextual grouping of the data streams for pre-processing by specialized Kalman filter units is found to be able to satisfy the logistics of IoT system operations. It is demonstrated that interconnection of the elementary Kalman filters into an organized network, the compositional Kalman filter, allows to take advantage of the redundancy of data streams to accomplish IoT pre-processing of the raw data. This includes intermittent data imputation, missing data replacement, lost data recovery, as well as error events detection and correction. Demonstrated is the efficiency of the suggested compositional designs of elementary Kalman filter networks for the purpose of data pre-processing in IoT systems.

In this paper, the difference between two oddball cases: Target with Response (TR) and Non-target with Response (NTR); is evaluated in terms of functional connectivity. For statistical significance Pearson's Correlation method is used. The connectivity was estimated using coherence on single trial EEG data from a visual oddball paradigm. The pairwise coherence coefficient of all 19 electrodes was grouped into a 19×19 connectivity matrix for further analysis. The data of two subj ects was averaged to perform group comparisons between the two oddball cases. Absolute thresholding method was applied on the matrices to form undirected graphs. It is found that the connectivity matrix of TR is significantly different from NTR. Firstly the connectivity in TR case has higher weight than NTR case. Secondly TR case has higher network density value than NTR case. Lastly Frontal-Temporal (F - T) and Occipital-Temporal (O-T) links which are related to the triggering of P300 were observed in TR case but not in NTR case. It is also observed that Central-Parietal (C-P) and Parietal-Temporal (P-T) links are important for the motor response.

In this paper, a video based contactless heart rate monitoring system for a person driving a car is proposed. This system uses photoplethysmography (PPG) signal extracted from subject's face to measure his heart rate. The PPG signal acquired is effected by the illumination variation and motion artifacts that are induced when the car is moving in real life scenario. Hence, a series of filtering algorithms are applied to reduce the noise to obtain accurate heart rate. The video of subject's face is recorded for ten seconds using dashcam at a frame rate of $30fps$. The resolution of image is $640x480$ pixels. In each video frame, the subject's face is detected using Viola-Jones face detector algorithm and region of interest (ROI) is segmented to compute the average Red-Green-Blue (RGB) values. The raw PPG signal is then filtered using a series of algorithms such as signal detrending, signal normalization, illumination variation reduction, bandpass filtering, signal smoothing and Joint Approximate Diagonalization Eigenmatrices (JADE) Independent Component Analysis (ICA). Fast Fourier Transform (FFT) is used to transform the filtered PPG signal into frequency domain for peak detection. The frequency component that corresponds to the peak amplitude is the heart rate of the subject, measured in beats per minute (bpm).

Copper Pillar bump Flipchip in the key solution in providing the breakthrough in package size reduction, lower cost package and better product performance. However there are challenges in creating a reliable copper pillar bump. Proper choice of electroplating materials and process are critical in meeting the design goals, quality and reliability. During a qualification run on a copper pillar bump flipchip on a thin small no lead package, excessive Kirkendall voids growth was found in the copper and SnAg (Tin-sliver) layer after 1000 hours of High Temperature storage at 150 deg C and 1000 cycles temperature cycle from −55°C to 150°. Methodologies like 8D, fishbone diagram and 5 why are used to investigate the root cause of this abnormality formation and deriving a robust solution to resolve the issue. The Investigation include evaluation of 2 different plating chemical and the plating parameter optimization with some improvement on the current copper pillar bump stack up. The evaluation will further validate by HTSL 175 deg C for 168 hours and X-ray and X-sectional analysis of packages.