DR-NTU (Digital Repository of NTU)
Not a member yet
116018 research outputs found
Sort by
BrainOOD: out-of-distribution generalizable brain network analysis
No full textIn neuroscience, identifying distinct patterns linked to neurological disorders, such as Alzheimer’s and Autism, is critical for early diagnosis and effective intervention. Graph Neural Networks (GNNs) have shown promising in analyzing brain networks, but there are two major challenges in using GNNs: (1) distribution shifts in multi-site brain network data, leading to poor Out-of-Distribution (OOD) generalization, and (2) limited interpretability in identifying key brain regions critical to neurological disorders. Existing graph OOD methods, while effective in other domains, struggle with the unique characteristics of brain networks. To bridge these gaps, we introduce BrainOOD, a novel framework tailored for brain networks that enhances GNNs’ OOD generalization and interpretability. BrainOOD framework consists of a feature selector and a structure extractor, which incorporates various auxiliary losses including an improved Graph Information Bottleneck (GIB) objective to recover causal
subgraphs. By aligning structure selection across brain networks and filtering noisy features, BrainOOD offers reliable interpretations of critical brain regions. Our approach outperforms 16 existing methods and improves generalization to OOD subjects by up to 8.5%. Case studies highlight the scientific validity of the patterns extracted, which aligns with the findings in known neuroscience literature. We also propose the first OOD brain network benchmark, which provides a foundation for future research in this field. Our code is available at
https://github.com/AngusMonroe/BrainOOD.Ministry of Education (MOE)Published versionThis research/project is supported by the Ministry of Education, Singapore under its MOE Academic Research Fund Tier 2 (STEM RIE2025 Award MOE-T2EP20220-0006) and Tier 1 (RG16/24). Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not reflect the views of the Ministry of Education, Singapore. YQ and JC are supported by Research Grants 8601116, 8601594, and 8601625 from the UGC of Hong Kong
Learning Arabic language: motivational factors of students in Singapore
No full textRising awareness of the value of learning a foreign language has led to many learning foreign languages for various reasons. While globalisation has motivated many to learn one or two foreign languages to enhance their skills and capability in various aspects, such as business, society, education, politics, and society, some are driven by a sense of personal achievement that they want to attain. This paper examines the motivational factors influencing students at Nanyang Technological University to learn Arabic as a foreign language.
This study aims to identify the intrinsic and extrinsic factors influencing students’ motivation and achievement in learning a foreign language. It examines aspects such as demographics, fields of study, and the underlying reasons for enrolling in language courses. The researcher adopts a quantitative approach, utilizing survey analysis to gather responses from 45 students across various years of study, specialisations and faculties. The results reveal a combination of intrinsic and extrinsic motivational factors influencing students’ engagement in foreign language learning, encompassing accomplishment factors, knowledge factors, religious factors, and expected learning outcomes. The findings will help more educators, curriculum developers, and language institutes in teaching and developing their syllabi.Published versio
3iGS: factorised tensorial illumination for 3D Gaussian splatting
No full textThe use of 3D Gaussians as representation of radiance fields has enabled high quality novel view synthesis at real-time rendering speed. However, the choice of optimising the outgoing radiance of each Gaussian independently as spherical harmonics results in unsatisfactory view dependent effects. In response to these limitations, our work, Factorised Tensorial Illumination for 3D Gaussian Splatting, or 3iGS, improves upon 3D Gaussian Splatting (3DGS) rendering quality. Instead of optimising a single outgoing radiance parameter, 3iGS enhances 3DGS view-dependent effects by expressing the outgoing radiance as a function of a local illumination field and Bidirectional Reflectance Distribution Function (BRDF) features. We optimise a continuous incident illumination field through a Tensorial Factorisation representation, while separately fine-tuning the BRDF features of each 3D Gaussian relative to this illumination field. Our methodology significantly enhances the rendering quality of specular view-dependent effects of 3DGS, while maintaining rapid training and rendering speeds.Nanyang Technological UniversityThis study is supported under the RIE2020 Industry Alignment Fund - Industry Collaboration Projects (IAF-ICP) Funding initiative, as well as cash and in-kind collaboration from the industry partner(s)
Investigating alternative signaling pathways regulated by Kindlin-3
No full textA three-membered, FERM-domain containing, focal adhesion protein family known as
Kindlin mediates cellular adhesion and migration by promoting Integrin activation. This
protein family comprises Kindlin-1, Kindlin-2 and Kindlin-3. Both Kindlin-1 and Kindlin-2
mediate the activation of various signaling pathways, whether Kindlin-3 plays a similar role
remains to be studied. In this thesis, we examine the role Kindlin-3 plays in regulating
alternative signaling pathways. Using PTEN-deficient T-ALL cells, we show that Kindlin-3
facilitates MAPK/ERK activation to promote cell proliferation by mediating DDR.
Mechanistically, the Kindlin-3-RACK1-MEK1/2 ternary complex is transported to the PIP3-
rich plasma membrane via Kindlin-3 binding to PIP3. Following PIP3 binding, RACK1-
MEK1/2 dissociates from Kindlin-3, leading to RACK1 scaffolding MEK1/2 and b-RAF.
This brings the MAPK/ERK components in proximity to promote MAPK/ERK activation.
Ablation of Kindlin-3 expression leads to diminished ERK1/2 activation, elevated levels of
DNA damage and delayed mitotic completion. Interestingly, the Kindlin-3 splice isoform,
Kindlin-3L is unable to facilitate MAPK/ERK activation due to its poor PIP3-binding ability.
During TCR activation, PIP3 production and Integrin LFA-1 activation promotes Kindlin-3
plasma membrane recruitment, contributing to MAPK/ERK activation. We also provide
preliminary evidence that Kindlin-3 interacts with the cytokine receptor, IL2Rβ to facilitate
its recruitment to the T-T homotypic synapse. Ablation of Kindlin-3 expression leads to
diminished IL2Rβ levels at the T-T homotypic synapse. We identified the binding interface
to be formed by the Kindlin-3 F0 subdomain and the C-terminal 21 amino acid stretch found
within the IL2Rβ STAT binding region.Doctor of Philosoph
Towards efficient and certified recovery from poisoning attacks in federated learning
No full textFederated learning (FL) is vulnerable to poisoning attacks, where malicious clients manipulate their updates to affect the global model. Although various methods exist for detecting those clients in FL, identifying malicious clients requires sufficient model updates, and hence by the time malicious clients are detected, FL models have been already poisoned. Thus, a method is needed to recover an accurate global model after malicious clients are identified. Current recovery methods rely on (i) all historical information from participating FL clients and (ii) the initial model unaffected by the malicious clients, both leading to a high demand for storage and computational resources. In this paper, we show that highly effective recovery can still be achieved based on (i) selective historical information rather than all historical information and (ii) a historical model that has not been significantly affected by malicious clients rather than the initial model. In this scenario, we can accelerate the recovery speed and decrease memory consumption as well as maintaining comparable recovery performance. Following this concept, we introduce Crab (Certified Recovery from Poisoning Attacks and Breaches), an efficient and certified recovery method, which relies on selective information storage and adaptive model rollback. Theoretically, we demonstrate that the difference between the global model recovered by Crab and the one recovered by train-from-scratch can be bounded under certain assumptions. Our experiments, performed across four datasets with multiple machine learning models and aggregation methods, involving both untargeted and targeted poisoning attacks, demonstrate that Crab is not only accurate and efficient but also consistently outperforms previous approaches in recovery speed and memory consumption.Info-communications Media Development Authority (IMDA)Ministry of Education (MOE)National Research Foundation (NRF)Submitted/Accepted versionThis research is supported by the National Research Foundation, Singapore and Infocomm Media Development Authority under its Trust Tech Funding Initiative, and the Singapore Ministry of Education Academic Research Fund (RG91/22 and NTU startup)
Novel reconfigurable battery energy storage system
No full textBattery energy storage systems (BESSs) are widely utilized in various applications, e.g., portable electrical devices, electric vehicles, and microgrids. However, conventional BESS structure with multiple battery cell/module/packs and fixed configuration exhibits the following deficiencies: adaptiveness, fault tolerance, and battery imbalance issue, which severely affects BESS reliability, capacity utilization, and battery lifespan. To tackle these issues, the reconfigurable BESSs are explored as potential solution in this thesis. These reconfigurable BESSs provides reconfiguration flexibility, fault tolerance capability and individual battery level active controllability. Therefore, this thesis proposes a modular reconfigurable BESS (MR-BESS) with improved characteristics and develops an equalizer-free balancing technique for reconfigurable type BESSs to bridge aforementioned gaps. The theoretical analysis of the proposed MR-BESS and balancing algorithm are supported with experimental results on hardware prototype to validate the effectiveness.Doctor of Philosoph
Battling falsehoods on China's short video platforms
No full textThe rapid rise of short video platforms in China has created a fertile ground for creating and spreading falsehoods. This commentary explores the nature of falsehoods on these platforms, examines existing countermeasures, and discusses the challenges, barriers, and potential solutions.Published versio
The time in thermal time
No full textPreparing general relativity for quantization in the Hamiltonian approach leads to the ‘problem of time’, rendering the world fundamentally timeless. One proposed solution is the ‘thermal time hypothesis’, which defines time in terms of states representing systems in thermal equilibrium. On this view, time is supposed to emerge thermodynamically even in a fundamentally timeless context. Here, I develop the worry that the thermal time hypothesis requires dynamics—and hence time—to get off the ground, thereby running into worries of circularity.Submitted/Accepted versio
Event-triggered differential delay method for current sharing in islanded DC microgrids with adaptive droop coefficient regulation
No full textIn islanded DC microgrids, various parallel distributed generation units (DGUs) are in voltage control mode, and mismatched line impedances always lead to inaccurate current sharing, which influences the normal and effective operation of power systems. To solve this problem, a distributed event-triggered differential delay method is proposed for current sharing in islanded DC microgrids. In this method, the master DGU is selected with the differential delay method, and then broadcasts its per-unit current via the communication network. When other DGUs have received this information, droop coefficients are adaptively regulated based on per-unit current errors, with which line impedance mismatches are eliminated. Furthermore, the communication and current sharing loops are activated with event-triggered control. With the proposed strategy, the current both in steady states and during load change transients is achieved with small droop coefficients, the bus voltage quality is hardly degraded. Besides, with the proposed event-triggered differential delay mechanism, both the date amount in communication and data exchange frequency are dramatically reduced. The convergence is analyzed and on this basis, the parameters in the triggering function are selected. Finally, hardware-in-loop test results in various conditions are presented to validate the effectiveness of the proposed strategy.This work was supported in part by the National Natural Science Foundation of China under Grant 52272403, and in part by the Aeronautical Science Foundation of China under Grant 20230028053003
Broadening our shared spaces through intercultural dialogue
No full textIn an increasingly fragmented world, the importance of dialogue in bridging divides cannot be understated. Indeed, interfaith and inter-ethnic dialogue has been consistently touted as the primary means of promoting understanding and acceptance among different communities in Singapore. However, as our identities and affiliations extend beyond race and religion, we must consider whether our efforts at dialogue should become more inclusive and all-encompassing. Instead of speaking merely of interfaith and inter-ethnic dialogue, it is possible to complement this with an intercultural lens that would allow us to cover any potential blind spots.Published versio