DR-NTU (Digital Repository of NTU)
Not a member yet
116018 research outputs found
Sort by
Managing senior management effectively - a study of essential success factors for project managers
No full textIn this fast-paced day and age of project management, project managers increasingly are working on projects that are volatile, uncertain, complex and ambiguous in nature. Coupled with the reality that project managers are in the sandwiched level of any given organisation, they often have to juggle between a wide array of project responsibilities as well as managing multiple cross functional senior management. This often results in project managers failing to effectively manage senior management which might lead to adverse project outcomes. From the author’s personal experiences, the challenges briefly shared are just the tip of the iceberg as there are often numerous root causes with multiple contributory factors to different contexts. Therein lies the core objective and motivation for this dissertation, which is for project managers to avoid preventable adverse project outcomes through effective senior management engagement. Recognizing that senior management have the power to positively or negatively influence project outcomes, the literature review will analyse the published literature of the identified domains of the dissertation, which would form the theoretical basis that would guide the research. Through a narrative inquiry approach, the author seeks to validate the domains of the literature review through collecting data from personal experiences of project managers and senior management. Additionally, challenges faced by project managers and senior management in the topic’s context would be analysed to determine commonalities which would serve as the foundation for the recommendations of this dissertation. The intent of these recommendations is to be versatile such that project managers from any industry and cultural context are able to utilize to facilitate successful influence of senior management to enhance project outcomes and success. This is achieved through the creation of a framework that was synthesized from effective approaches and solutions for project managers to effectively manage the demanding varying needs of direct and cross functional senior management.Master's degre
乡土·生态：《山南水北》中的“复魅”书写 = Soil and ecology: the “re-enchantment” in the writing of Shan Nan Shui Bei
No full text二十世纪以来，中国步入工业化进程，经济建设也逐渐变成国家发展的核心目标。然而在快速发展的同时，自然资源遭到过度开采，从而导致森林砍伐、土壤侵蚀、水资源浪费等环境问题。这些问题由于在当时并未得到足够重视，因此随着城市化、现代化的激进推进反而愈演愈烈。在这样的背景下，乡村成为了首当其冲的受害者，进而成为了许多作家笔下的书写对象。当代中国重要作家之一韩少功亦不例外，他在跨文体长篇作品《山南水北》中，展现了工业化和城市化对乡村生态的冲击，同时也彰显了其生态意识与乡土情怀。
本文首先回顾“乡土文学”和“生态文学”的概念，并对二者结合而产生的“乡土生态文学”进行阐释。之后，再对“乡土生态文学”与《山南水北》之间的内在关系展开讨论。其次，本文将界定“魅”的概念，并对马克斯·韦伯的“祛魅”理论与大卫·雷·格里芬的“复魅”理论进行阐释。之后，本文以格里芬的理论为基础，对《山南水北》的“复魅”美学展开分析。最终，本文得出结论：《山南水北》是乡土文学与生态文学的融合下，运用“复魅”美学构建独特的生态叙事与文化反思的作品。其不仅拓展了生态文学的表达维度，也为当代文学提供了新的思考路径。
Since the 20th century, China has embarked on a process of industrialization, with economic development gradually becoming the central focus of national progress. However, rapid growth has come at the cost of excessive resource exploitation, leading to environmental issues such as deforestation, soil erosion, and water resource depletion. These problems were not sufficiently addressed at the time. As urbanization and modernization advanced aggressively, the problems became increasingly severe. Under this context, rural areas bore the brunt of the impact and became a focal point in the writings of many authors. Han Shaogong, one of the most significant contemporary Chinese writers, is no exception. In his cross-genre novel Shan Nan Shui Bei, he not only depicts the impact of industrialization and urbanization on rural areas but also conveys his ecological consciousness and deep affection for rural life.
This paper first reviews the concepts of “Native Soil Literature” and “Ecological Literature”, as well as elaborates on the fusion of these two into “Native Ecological Literature”. Then, the paper will explore the relationship between this new genre and Shan Nan Shui Bei. Secondly, the paper defines the concept of “enchantment” and discusses Max Weber’s theory of “disenchantment” and David Ray Griffin’s theory of “re-enchantment”. Based on Griffin’s theory, this paper then analyzes the aesthetics of “re-enchantment” in Shan Nan Shui Bei. Finally, the paper concludes that Shan Nan Shui Bei is a work that integrates native soil literature and ecological literature, employing the aesthetics of “re-enchantment” to construct a unique ecological narrative and cultural reflection. Not only does it expand the expressive dimensions of ecological literature, but it also offers new perspectives for contemporary literary studies.Bachelor's degre
Head movement mimic - video-based human head motion transfer
No full textIn an era where deepfake technology continues to evolve, human head motion transfer
represents an innovative application with vast potential in entertainment and virtual
reality. This project aims to develop a web application that leverages the Thin-PlateSpline (TPS) Motion Model, Motion Refinement for Unsupervised Face Animation
(MRFA) and Facial Scene Representation Transformer (FSRT) to map head movements
from a driving video onto a static image, generating a realistic mimicked video. By
incorporating principles of Human-Computer Interaction (HCI), the application is
designed for ease of use and accessibility, allowing non-technical users to engage with
cutting-edge technologies. This report discusses the system’s design, implementation,
and user studies conducted to evaluate usability. Additionally, the project integrated
the models, and evaluated usability of the application through testing with a diverse
set of participants. The findings demonstrated the potential of these models in motion
transfer applications and the importance of balancing technical performance with user
experience.Bachelor's degre
Controllable image generation using diffusion models
No full textThis research investigates advanced control mechanisms for diffusion-based image
generation models, focusing on developing a comprehensive pipeline for highly controllable
outputs. The study evaluates four key control approaches: prompt engineering, ControlNet
guidance, Low-Rank Adaptation (LoRA), and inpainting techniques. Through systematic
experimentation with the Flux.dev diffusion model, the research demonstrates significant
improvements in character consistency, pose control, and targeted image refinement.
The findings reveal that while prompt engineering provides effective stylistic guidance, it
faces limitations in spatial relationship control and character consistency. LoRA training
offers a resource-efficient alternative to full model fine-tuning, though with slightly reduced
accuracy. ControlNet integration proves particularly effective for structural and pose
guidance, with different pre-processors showing distinct strengths in various applications.
The research culminates in a unified pipeline that combines these mechanisms, demonstrating
practical applications in content creation, professional photography, and e-commerce.
The study's outcomes provide valuable insights for implementing controllable image
generation in real-world applications while identifying areas for future technical enhancement
and research directions.Bachelor's degre
After the tone
No full textAfter The Tone is a 14-minute documentary about grief and bereavement. As a topic that is often swept under the carpet, we have organised this film as a biography of grief through the lens of love, without shying away from the non-mutually exclusive pain it brings. We feature three profiles who have each lost a loved one. They bring us through their grief, in relation to their own lives as well as in relation to society – namely, interactions with others who do not know how to properly express their well-intended support, and the societal pressure to move on quickly. Through this film, we hope to impart to the audience a heightened level of grief literacy, and an informed version of empathy for the bereaved.Bachelor's degre
Exploring the role of twist angle in RESE2-HBN heterostructures for optoelectronic
No full textIn recent years, two-dimensional materials have attracted much attention due to their unique electronic, optical and mechanical properties. Among these materials, RESE2, as an emerging material, has attracted attention for its torsion angle-dependent properties exhibited in heterostructures. The aim of this study is to investigate the photoluminescence properties of RESE2-HBN heterostructures, focusing on the relationship between torsion angle modulation and exciton behavior.
In this study, we first obtained suitable RESE2 and HBN samples with sizes of about 5µm-10 µm and 15µm-20µm, respectively, by mechanical stripping technique, and then assembled RESE2 and HBN stacked according to the torsion angles of 2°, 4°, and 6° to form high-quality heterostructures by using a precise PC transfer technique. During the transfer process, the integrity of the samples and the quality of interlayer coupling were ensured by optimizing the transfer temperature and pressure. Subsequently, the prepared samples were tested for photoluminescence spectroscopy at 10 K cryogenic temperature. The experimental data were fitted with Lorentzian functions to reveal the remarkable features of excitons and triplet excitons and their patterns with torsion angle.
This study provides new insights into the tunable optical properties of RESE2-HBN heterostructures and opens new avenues for designing advanced optoelectronic devices. By introducing the torsion angle modulation technique, the practical applications of 2D materials in the fields of optoelectronic sensing, quantum computing and energy conversion can be further expanded. Future research could focus on achieving higher precision angular control and exploring heterogeneous integration schemes for other 2D materials to further enhance functionality and application potential.Master's degre
Multi-band rectenna design for wireless power transmission
No full textWith the rapid development of miniaturization of electronic equipment and wireless power transmission, the collection of ambient microwave energy has received widespread attention in the industry, where a rectenna is responsible for collecting electromagnetic energy and converting it into DC energy. This
dissertation designs an ultra-wideband receiving antenna and rectifier circuit with a working frequency band of 0.9-3.5 GHz for the GSM-900, WiFi, and WiMAX bands that are widely used in communication bands.
Firstly, theoretical basis concerning maximum power transfer efficiency and extended kQ product is introduced, together with derivation of expression relationship between different variables in terms of Rayleigh quotient and
eigenvalue. An example of inductive power transfer system with specific values is introduced to verify derived formula. Then an Archimedean planar spiral antenna is designed and simulated with a index gradient balun structure to feed the spiral antenna, and simulation results show that in the target frequency band of 0.9-3.5 GHz, the S11 is below -10 dB, the axial ratio is below 3 dB, and the antenna gains at 0.9 GHz, 2.45 GHz
and 3.5 GHz frequency points are 3.8 dBi, 7.5 dBi and 4.4 dBi respectively. Two types of rectifier circuits are designed: one is a wideband rectifier circuit with an operating frequency band of 0.8-1.0 GHz, and a Chebyshev bandpass input filter is designed to improve the performance; the other is a dual-band rectifier circuit with an operating frequency of 2.45 GHz and 3.5 GHz. On
this basis, the two rectifier branches are integrated and simulation results show
that S11 keeps below -20 dB at the target frequency when the input power is 14 dBm and the DC load is 500 Ω. The rectification efficiency at the three frequencies reaches 13.1%, 12.5% and 16.5%, and the output DC voltage is 1.28 V, 1.25 V and 1.44 V respectively. which is capable to realize the
function of collecting ambient microwave energy in space.Master's degre
Rewriting the hybrid: sinification and identity negotiations under Hong Kong's national security law regime
No full textThis study investigates how state authorities and members of the Hong Kong public negotiate between a distinct Hong Kong identity and a Chinese identity under the current National Security Law regime, and the implications it has for the city’s autonomy. Employing a mixed-methods research design, three critical spaces for identity negotiations are analyzed during the timeframe of June 2020 through March 2025: legal, educational, and transnational. Drawing on Bhabha’s concept of hybridity and Gramsci’s cultural hegemony, it argues that Beijing has increased its presence in local Hong Kong institutions through coercive and consensus-based promotion of a China-centric identity. In the legal space, trials are used to delegitimize Hongkonger values tied to civil liberties, while patriotic education rewrites history, morality and identity to align Hong Kong people with the Mainland. Through online and diasporic activities, the Hongkonger identity continues to be expressed, drawing on both anti-Chinese sentiments and addressing its colonial legacy. However, with limited counter-hegemonic ability inside the city, the Hong Kong hybrid identity is increasingly becoming sinified, reducing the strength of values behind protests of both 2014 and 2019, tightening Beijing’s grip of the city.Master's degre
RSA and its weak implementations
No full textRSA is a public key cryptosystem which hardness is linked to two problems — the RSA Problem and the Integer Factorisation Problem. While the two aforementioned problems may be difficult to solve when large RSA moduli are used, there may be shortcuts in solving these problems given a weak implementation of RSA. A weak implementation may arise from implementing one’s own RSA instead of using well-maintained software libraries for cryptography. Examples of weak implementations include inadequate primes p and q, and the improper selection of public and private exponents e and d.
In this paper, various weak implementations of RSA were explored and possible attacks for such weak implementations were explained and demonstrated on real-life RSA modulus sizes (i.e., 2048 and 3072 bits). The attacks were split into two categories. The e-th root attack on the use of small e in conjunction with short messages as well as Håstad’s Broadcast attack belonged to the first category of attacks that result in direct plaintext recovery. The attack on neighbouring primes, Wiener’s attack on small d and the Coppersmith’s attack on the partial leakage of primes belonged to the second category of attacks that result in the recovery of primes p and q or the private exponent d.
The successful demonstration of the attacks on weak implementations on real-life RSA modulus sizes highlighted the potential dangers of implementing one’s own RSA, which would not be as thoroughly-tested to identify underlying weaknesses. While well-known and well-maintained cryptographic libraries may not be immune to future attacks, there is certainly less chance of weak implementations.Bachelor's degre
Human-centric 3D representation learning
No full textUnderstanding the world in three dimensions has long been a scientific challenge with significant practical implications. In the era of deep learning, researchers have made substantial progress in developing 3D representations using deep neural networks. Among the myriad of entities in our complex environment, humans stand out due to their general significance in science and their specific relevance in various applications. This thesis focuses on human-centric 3D representation learning within the framework of deep learning. Specifically, we explore three key areas: human-centric 3D perception, human reconstruction, and 3D human generation. The thesis presents five studies that collectively address these topics.
In the realm of human-centric 3D perception, we introduce a versatile multi-modal pre-training approach. By harnessing the diverse modalities of human data, such as RGB images, depth, and 2D keypoints, we present a general framework HCMoCo for effective human-centric representation learning. This framework achieves state-of-the-art performance across four human perception tasks, including DensePose prediction, human parsing, and 3D keypoint prediction.
In the area of human reconstruction, we investigate garment reconstruction from 4D point clouds of dressed individuals. To address the ambiguities inherent in 2D images, we propose a principled framework called Garment4D, which enables separable and interpretable garment reconstruction. Notably, Garment4D can effectively reconstruct and model the non-rigid deformations of loose garments (e.g., skirts) that do not share the same topology as the human body.
We present two distinct approaches to 3D human generation. In our first work, AvatarCLIP, we generate and animate avatars based on text descriptions of body shapes, appearances, and motions. By leveraging differentiable rendering and large- scale vision-language pre-trained models, we achieve avatar and motion synthesis without the need for supervised training or paired data. To further enhance the quality of the generated avatars, we introduce a second approach, EVA3D, which is a high-quality unconditional 3D human generative model that requires only 2D image collections for training. We design an efficient compositional human NeRF representation to facilitate high-resolution 3D human sampling and rendering, which is employed in adversarial training.
Finally, by harnessing the capabilities of large language models (LLMs), we explore 3D human representation learning with a focus on human motion and the integration of perception, reconstruction, and generation. We introduce EgoLM, a versatile framework for understanding egocentric motion using multi-modal data. This framework incorporates rich contextual information from egocentric videos and motion sensors provided by wearable devices. EgoLM unifies various motion learning tasks, including motion understanding from video and motion data, as well as motion tracking and generation from text or sparse sensor input. Additionally, it enables a novel task unique to wearable devices: generating text descriptions from sparse sensor data.
By investigating 3D human representation learning from three distinct perspectives, we have attained a comprehensive understanding of humans that spans both high-level semantics and low-level geometry. This thesis offers a cohesive exploration of human-centric 3D representations, contributing significantly to the field of human-centric vision.Doctor of Philosoph