DR-NTU (Digital Repository of NTU)
Not a member yet
116018 research outputs found
Sort by
Too beautiful to be fake: attractive faces are less likely to be judged as artificially generated
No full textTechnological advances render the distinction between artificial (e.g., computer-generated faces) and real stimuli increasingly difficult, yet the factors driving our beliefs regarding the nature of ambiguous stimuli remain largely unknown. In this study, 150 participants rated 109 pictures of faces on 4 characteristics (attractiveness, beauty, trustworthiness, familiarity). The stimuli were then presented again with the new information that some of them were AI-generated, and participants had to rate each image according to whether they believed them to be real or fake. Despite all images being pictures of real faces from the same database, most participants did indeed rate a large portion of them as 'fake' (often with high confidence), with strong intra- and inter-individual variability. Our results suggest a gender-dependent role of attractiveness on reality judgements, with faces rated as more attractive being classified as more real. We also report links between reality beliefs tendencies and dispositional traits such as narcissism and paranoid ideation.Nanyang Technological UniversityPublished versionThis work was supported by the Presidential Postdoctoral Fellowship Grant (NTU-PPF-2020-10014) from Nanyang Technological University (awarded to DM)
Local polar order controls mechanical stress and triggers layer formation in Myxococcus xanthus colonies
No full textColonies of the social bacterium Myxococcus xanthus go through a morphological transition from a thin colony of cells to three-dimensional droplet-like fruiting bodies as a strategy to survive starvation. The biological pathways that control the decision to form a fruiting body have been studied extensively. However, the mechanical events that trigger the creation of multiple cell layers and give rise to droplet formation remain poorly understood. By measuring cell orientation, velocity, polarity, and force with cell-scale resolution, we reveal a stochastic local polar order in addition to the more obvious nematic order. Average cell velocity and active force at topological defects agree with predictions from active nematic theory, but their fluctuations are substantially larger than the mean due to polar active forces generated by the self-propelled rod-shaped cells. We find that M. xanthus cells adjust their reversal frequency to tune the magnitude of this local polar order, which in turn controls the mechanical stresses and triggers layer formation in the colonies.Nanyang Technological UniversityPublished versionThis work was supported by the NSF through awards PHY-1806501 and PHY-2210346, and the Center for the Physics of Biological Function (PHY-1734030). E.H. acknowledges the NTU Start-Up Grant. R.A. acknowledges funding from the European Union through the ERC Starting Grant “Living Fluctuations” (No. 101114584). N.S.W. acknowledges National Institutes of Health Grant R01 GM082938
The influences of adopting a healthy lifestyle in young adult women
No full textThe health lifestyles of young adult women are becoming increasingly important as health issues are often interlinked and the present lifestyle habits are found to have consequences on quality of later life. While previous research on healthy lifestyle habits has encompassed other broad groups of individuals, little research is conducted on the influences of health lifestyles of young adult women. As such, this study seeks to investigate the influences of culture, gender, and socioeconomic status in adopting a healthy lifestyle in young adult women. Using 12 semi-structured interviews, this study explores the lifestyle habits of young women, their perceptions of them, and the various influences in their lifestyle habits, specifically in terms of exercise and dietary habits. Findings from this study highlight structural and cultural facilitators, barriers, and motivators of healthy lifestyle habits. Underlying attitudes, subjective norms and perceived control over lifestyle habits were also found to be important in pursuing healthier lifestyle behavior.Bachelor's degre
Conductive-bridge resistive random access memory
No full textThe gap between the requirement to process data and an organization's available
bandwidth constitutes today's Big Data Problem [1]. In a Data-Driven Economy, data
with Volume, Velocity, Variety, and Veracity is an asset to any organization [1,2].
Memory must meet the high-density technological demands of the market [1,7].
The existing Von-Neumann architecture separates a system's computational and
memory-storing processes. Hence, instructions require back-and-forth data
transportation in large volumes, incurring latency and energy costs that effectively
bottleneck the progress of data processing capabilities sought in the market [6,7]. To
fulfil the high-speed memory demands that drive the big data problem, an alternative
approach can fill the technological gap. Neuromorphic computing was introduced,
where computation occurs within the memory component, and there currently exists a
need to find new technologies that can fulfil a storage class memory and address the
existing hurdles faced in memory. [3,6,7]
This paper will provide insight into the Cycle-to-Cycle(C-C) and Device-to-Device
(D-D) variability of multiple CBRAM devices that consist of a Pt-GeTe-Ag and a PtGeS-Ti-Ag CBRAM device with a Metal Insulator Metal (MIM) Structure. Pt-GeTeAg demonstrated SET and RESET on the opposite polarity and initially showed
promise for analysis over 100 cycle endurance. Still, fabrication issues prevented
reproducible results, which held back research progress. SET and RESET distributions
and measurements of the HRS and LRS of Pt-GeS-Ti-Ag were plotted. Pt-GeS-Ti-Ag
shows the potential to display satisfactory endurance and variability. Future research
should continue exploring CBRAM with GeS over GeTe.Bachelor's degre
More than cute
No full text"More than Cute" is a project that aims to educate the public on the responsibilities of small pet ownership and dispel harmful myths. By exploring the space, diet, social, and enrichment needs of these animals, the project seeks to foster empathy and promote proper care practices. Through engaging design collaterals, including a general pamphlet and detailed booklets for each pet, this initiative hopes to inspire thoughtful and informed decisions, emphasizing that small pets deserve the same level of care and respect as larger animals.Bachelor's degre
Connecting dissimilar optical fibres
No full textEfficient coupling between single-mode fibers and large mode area fibers remains a key challenge in high-power fiber lasers and advanced optical communication systems, due to significant mismatches in mode field diameter and numerical aperture. This thesis explores the use of an intermediate fiber to enhance coupling efficiency between a standard SMF-28 and a specific LMA fiber model (IXF-2CF-Yb-O-20-130-NL, which represents an industrial standard). A simulation framework is built in MATLAB to study the impact of the intermediate fiber’s MFD and NA on total coupling efficiency.
The simulation considers LP01 and LP11 modes, modeled by Gaussian and Bessel functions respectively. Coupling efficiency is computed using the mode overlap integral method for LP01→LP01 and LP01→ LP11 transitions, and a weighted total is calculated assuming a power ratio of 0.95:0.05.
Results demonstrate that the optimized intermediate fiber achieves a maximum efficiency of 0.7603 under ideal alignment, outperforming both direct coupling (0.6402) and commercial mode field adapters (0.7304). The study provides a practical and adaptable solution for improving fiber-to-fiber coupling in multimode systems.Master's degre
Compressive behaviour of steel open section stub columns fabricated by 3D printing (WAAM)
No full textThis report presents an experimental investigation into the structural behaviour of wire arc
additive manufactured (WAAM) stainless steel of Grade 316L. A total of nine I-shaped and
cruciform-shaped stub column specimens were tested under axial compression. The relevant
testing machine, specimen setup, and test procedures were reported, together with the resulting
load-displacement curves, ultimate capacities, and failure modes analysed. Due to the absence
of design codes on wire arc additive manufactured stainless steel, experimental results were
compared against three international design standards: (i) EN 1993-1-3 (Eurocode 3), (ii) AISI
S100 (North American Specification), and (iii) AS/NZS 4600 (Australian/New Zealand
Standard). All codes provided conservative estimates, with Eurocode 3 showing the best
agreement. These findings suggested that while current design methods can be applied to
WAAM columns, further refinement is needed to account for unconventional geometries and
the unique material properties of WAAM stainless steel.Bachelor's degre
Research on high-speed power switching characteristics and application of gallium nitride HEMT
No full textWhile traditional Si devices struggle to meet the escalating performance
requirements of contemporary power electronics, wide-bandgap semiconductor
power devices have revolutionized switching capabilities by enabling
megahertz-level operation, substantially improving both energy efficiency and
system compactness. The semiconductor switch, as the critical component in power
converters, fundamentally determines the operational characteristics of these
systems. Gallium nitride (GaN) technology, distinguished by its exceptional material
advantages, is progressively displacing conventional Si solutions and catalyzing the
development of converters with enhanced frequency ranges, superior efficiency
metrics, and optimized power-to-volume ratios. Notably, the Cascode GaN HEMT
exemplifies this technological evolution. Its hybrid design integrates a low-voltage
Si MOSFET with a high-voltage depletion-mode GaN HEMT in series, offering
exceptional adaptability for medium/high-voltage implementations.
Building on these advancements, this research systematically explores the
high-frequency switching mechanisms and implementation strategies of cascode
GaN technology. The investigation begins with a comparative evaluation of static
performance parameters between cascode GaN and traditional Si devices, with
focused analysis on forward-blocking capability, conduction characteristics, and
reverse-mode operation. Given the critical influence of parasitic components under
high-speed switching scenarios, we developed a parasitic-inclusive double-pulse
testing methodology to decode transient switching phenomena. Subsequently, a
dedicated experimental setup was implemented to acquire high-fidelity switching
waveforms, enabling quantitative verification of theoretical models.
Building on this foundational analysis, we identify the principal factors
influencing the switching characteristics of cascode GaN devices. Through a
combination of theoretical derivations and circuit simulations, we evaluate the
effects of three categories of parameters — gate-driver parameters, operating
conditions, and parasitic inductances — on switching performance. From these
studies, we derive guidelines for selecting driver parameters and propose that
parasitic inductances be optimized according to the specific tradeoffs between
efficiency and reliability in the intended application. We further present targeted
mitigation schemes for overshoot and ringing phenomena and, by comparative
analysis, establish a hierarchy for parasitic-parameter optimization: source-loop
inductance, power-loop inductance, and driver-loop inductance, in that order.Master's degre
Disco4D: disentangled 4D human generation and animation from a single image
No full textWe present Disco4D, a novel Gaussian Splatting framework for 4D human generation and animation from a single image. Different from existing methods, Disco4D
distinctively disentangles clothings (with Gaussian models) from the human body
(with SMPL-X model), significantly enhancing the generation details and flexibility.
It has the following technical innovations. 1) Disco4D learns to efficiently fit the
clothing Gaussians over the SMPL-X Gaussians. 2) It adopts diffusion models to
enhance the 3D generation process, e.g., modeling occluded parts not visible in the
input image. 3) It learns an identity encoding for each clothing Gaussian to facilitate
the separation and extraction of clothing assets. Furthermore, Disco4D naturally
supports 4D human animation with vivid dynamics. Extensive experiments demonstrate the superiority of Disco4D on 4D human generation and animation tasks.
Our visualizations can be found in https://disco-4d.github.io/.Agency for Science, Technology and Research (A*STAR)AI SingaporeNational Research Foundation (NRF)Ministry of Education (MOE)Submitted/Accepted versionThis research/project is supported by the National Research Foundation, Singapore under its AI Singapore Programme(AISG Award No: AISG3- PhD-2023-08-049T). This study is supported by the Ministry of Education, Singapore, under its MOE AcRF Tier 2 (MOET2EP20221-0012, MOE-T2EP20223-0002), and under the RIE2020 Industry Alignment Fund – Industry Collaboration Projects (IAF-ICP) Funding Initiative, as well as cash and in-kind contribution from the industry partner(s)
Uncovering conserved sporopollenin associated genes via cross-species co-expression analysis
No full textGene connectivity score was validated as a novel and biologically meaningful metric
for identifying regulatory genes in biosynthetic pathways. We show that connectivity
reflects both co-expression strength and organ specificity, with high-scoring genes
enriched in tissues where the pathway is active. By integrating connectivity scores of
orthologs across 14 plant species, we developed a cross-species framework that
filters out lineage-specific false positives and prioritises conserved candidates.
Applied to the sporopollenin biosynthesis pathway, this approach identified 25 highconfidence genes, with preliminary functional validation of 2 genes. The method
presented is highly accessible due to open-source tools like Large-Scale
Transcriptomic Analysis Pipeline in Cloud (LSTrAP) and Two-Tier Ensemble
Aggregation – Gene co-expression network (TEA-GCN) and can be readily applied
to any organism or pathway of interest using only a small set of known biosynthetic
genesBachelor's degre