Singapore University of Technology and Design

As more users seek generative AI models to enhance work efficiency, generative AI and Model-as-a-Service will drive transformative changes and upgrades across all industries. However, when users utilize generative AI models provided by the service provider, they cannot be certain that the model’s quality matches the provider’s claims. Considering the need to protect intellectual property, the service provider will not disclose model details for user verification. To this end, we take the Internet of Vehicles as research background, proposing a zero knowledge model proof architecture based on UAVs. We also introduce a multi-agent reinforcement learning algorithm to optimize the verification process. In specific, we first propose a verification scheme for the key operations of generative adversarial networks based on non-interactive zero knowledge proof. The zero knowledge proof architecture ensures that model parameters cannot be stolen during the verification process. After that, we propose an age of verification (AoV) metric to ensure the timeliness and freshness of zero knowledge proof. We also construct a trade-off optimization problem between the energy consumption of UAV as a verifier and the AoV of edge servers as service providers, and transform the problem based on Lyapunov optimization theory. Following that, we propose an enhanced multi-agent proximal policy optimization algorithm to enable the collaborative verification of edge servers by multiple UAVs. The algorithm simulation results demonstrate that the reward value of our proposed algorithm is over 10% higher than that of the standard algorithm, with a faster and more stable overall convergence speed. Additionally, the zero knowledge proof performance test results indicate that the verification delay in our proposed architecture is less than 500ms during the verification phase, meeting practical requirements.

Altermagnetism, the third magnetic phase beyond ferromagnetism and antiferromagnetism, holds great promise for spintronics but also faces challenges in materials discovery and mechanism exploration. Here, through high‐throughput screening of 170 pentagonal 2D materials, 4 altermagnetic semiconductors (MnS2, CoS2, MnC2, and CoPSe) are identified with sizable spin splitting (109–172 meV). It is shown that MnS2 is the first‐of‐kind altermagnetic second‐order topological insulator (AMSOTI), as evidenced by a nontrivial real Chern number (νR = 1) and spin‐polarized corner state. Using ab inito quantum transport simulation, a MnS2‐based altermagnetic tunneling junction (AMTJ) is designed and a giant tunneling magnetoresistance (TMR) of 1.5 × 10⁵% at the Fermi level, which surpasses MF2 (M = Co and Ni)‐based AMTJ by three orders of magnitude. In addition, by applying a twisting angle of 90° to the AMTJ without altering the Néel${\rm N}\acute{\rm {e}}{\rm el}$ order, a significant TMR of 1.2 × 10⁷% as driven by the structural changes is observed. This work unveils pentagonal 2D materials as a compelling material platform for exploring the physics and device applications of altermagnets.

The protection of quantum gates from arbitrary single‐ and two‐qubit noises with properly designed dynamical decoupling (DD) pulses is explored. The proposed dynamical decoupling method is a concatenation of a sequence of pulses formed by σj,kx$\sigma ^{x}_{j,k}$, σjxσkx$\sigma _{j}^{x}\sigma _{k}^{x}$ with another sequence constructed by σj,kz$\sigma ^{z}_{j,k}$, σjzσkz$\sigma ^{z}_{j}\sigma ^{z}_{k}$. The concatenation of the two sequences results in desired pulses to fight against any single‐ and two‐qubit errors. The success of the method relies on the ability to adjust system parameters or interaction terms, which can be achieved in different physical systems, including trapped ions and superconducting qubits. Finallythe performance of the method is explored numerically with the above‐mentioned errors that are changing at any moment and show the preferred protection offered by the method. Therefore, the method is a timely step forward in preserving quantum gates at the level of physical qubits.

Extreme heat events, exacerbated by climate change, are intensified in cities by the urban heat island (UHI) effect. A primary cause of UHI is the replacement of natural lands with buildings and roads, trapping heat in the city. Singapore, a high-rise high-density city state in Southeast Asia, has taken actions to mitigate UHI including the expansion of urban green spaces. Increasing urban greenery to mitigate UHI is one of many strategies being tested in the emerging field of climate-responsive city planning. To justify UHI mitigation planning strategies to city stakeholders, however, requires an ability to estimate effectiveness and efficiency in measures. This case study implements a generative urban model to evaluate UHI across a range of urban density scenarios, testing the impacts of both green space and building design strategies in three stages: (i) urban generative modeling, (ii) UHI prediction simulation, and (iii) urban data analysis. The study conducts urban generative modeling for a 100-ha site, using a model created to specifically reflect Singapore’s development control guidelines. Across eight selected design parameters, an extensive set of design solutions (9,000) is obtained via a one-factor-at-a-time sampling method. The Urban Weather Generator tool is used to evaluate the UHI performance per solution. Design space exploration of the urban model and analysis of results identify best-performing UHI mitigation strategies, correlation among model parameters, and parameter significance. These results permit discussion of effective city planning and design strategies for UHI mitigation. Readers engaging with this case study will gain an understanding of the application of urban data analytics to climate-resilient city planning supported by generative urban models and UHI performance simulation.

Graphical models have exhibited their performance in numerous tasks ranging from biological analysis to recommender systems. However, graphical models with hub nodes are computationally difficult to fit, particularly when the dimension of the data is large. To efficiently estimate the hub graphical models, we introduce a two-phase algorithm. The proposed algorithm first generates a good initial point via a dual alternating direction method of multipliers (ADMM), and then warm starts a semismooth Newton (SSN) based augmented Lagrangian method (ALM) to compute a solution that is accurate enough for practical tasks. We fully excavate the sparsity structure of the generalized Jacobian arising from the hubs in the graphical models, which ensures that the algorithm can obtain a nice solution very efficiently. Comprehensive experiments on both synthetic data and real data show that it obviously outperforms the existing state-of-the-art algorithms. In particular, in some high dimensional tasks, it can save more than 70% of the execution time, meanwhile still achieves a high-quality estimation.

Neuromorphic computing has the potential to overcome limitations of traditional silicon technology in machine learning tasks. Recent advancements in large crossbar arrays and silicon-based asynchronous spiking neural networks have led to promising neuromorphic systems. However, developing compact parallel computing technology for integrating artificial neural networks into traditional hardware remains a challenge. Organic computational materials offer affordable, biocompatible neuromorphic devices with exceptional adjustability and energy-efficient switching. Here, the review investigates the advancements made in the development of organic neuromorphic devices. This review explores resistive switching mechanisms such as interface-regulated filament growth, molecular-electronic dynamics, nanowire-confined filament growth, and vacancy-assisted ion migration, while proposing methodologies to enhance state retention and conductance adjustment. The survey examines the challenges faced in implementing low-power neuromorphic computing, e.g., reducing device size and improving switching time. The review analyses the potential of these materials in adjustable, flexible, and low-power consumption applications, viz. biohybrid spiking circuits interacting with biological systems, systems that respond to specific events, robotics, intelligent agents, neuromorphic computing, neuromorphic bioelectronics, neuroscience, and other applications, and prospects of this technology.

Nowadays, cyberattacks are growing at an alarming rate, causing widespread havoc to the digital community. In particular, authentication attacks have become a dominant attack vector, allowing intruders to impersonate legitimate users and maliciously access resources. Traditional single-factor authentication (SFA) protocols, which rely on a single authentication factor are often insufficient to address the growing sophistication of modern cyberattacks. To address the shortcomings in SFA, multi-factor authentication (MFA) protocols have been widely adopted in recent years, raising the security bar against impostors and restricting unauthorized accesses. MFA enhances security by incorporating multiple authentication factors, such as knowledge-based (e.g., passwords), possession-based (e.g., tokens), and inherent-based factors (e.g., biometrics), among others. However, while MFA is generally considered more secure than SFA, it is not foolproof. Because, critical vulnerabilities may still arise due to design or implementation flaws in MFA protocols. These vulnerabilities are often overlooked by designers or users and remain undetected until exploited by attackers, potentially resulting in catastrophic consequences. Unfortunately, existing works failed to adequately analyze and identify most of such critical security flaws in MFA protocols. In this work, we systematically analyze the intricate design and construction of MFA protocols to uncover potential design-level security flaws. To this end, we first define eight security evaluation criteria that are essential to critically evaluate design-level security flaws of MFA protocols. These criteria are primarily derived from existing and newly introduced MFA security requirements. We then review a range of MFA protocols across various domains. Using our established evaluation criteria, we perform a systematic security analysis and evaluation of these protocols, particularly focusing on their design and construction. Ultimately, we uncover several security flaws in most of the MFA protocols evaluated. Due to space limitation, we select ten of those protocols for deeper security analysis and provide a detailed discussion of the respective flaws identified. Additionally, we devised relevant mitigation strategies for each of the flaws identified. We believe that our findings provide valuable insights to cybersecurity researchers and practitioners to help them addressing a wide range of security flaws in MFA protocols.

Localized high‐concentration electrolytes offer a potential solution for achieving uniform lithium deposition and a stable solid‐electrolyte interface in Lithium metal batteries. However, the use of highly concentrated salts or structure‐loaded diluents can result in significantly higher production costs and increased environmental burdens. Herein, a novel localized high‐concentration electrolyte is developed, comprising ultra‐low content (2% by mass) triethylammonium chloride as an electrolyte additive. The stable Lewis acid structure of the triethylammonium chloride molecule allows for the adsorption of numerous solvent molecules and TFSI⁻ anions, intensifying the electrostatic interactions between lithium ions and anions. The chloride ions introduced by TC, along with TFSI⁻ anions, integrate into the solvent sheath, forming a LiCl‐rich inorganic SEI and enhancing the electrochemical performance of the lithium metal anode. The improved Li||Li cell shows excellent cycling stability for over 500 h at 1 mA cm² with a 27 mV overpotential. This work provides insights into the impact of electrolyte additives on the electrode‐electrolyte interface and Li‐ion solvation, crucial for safer lithium metal battery development.

The future sixth-generation (6G) paradigm aims to seamlessly integrate communication and environmental sensing capabilities into a single radio signal, promising improved efficiency and cost-effectiveness through simultaneous data communications and environmental perception. At the core of this evolution, orthogonal frequency division multiplexing (OFDM) and its advanced waveforms emerge as pivotal for integrated sensing and communications (ISAC). This study introduces a concise and unified ISAC waveform design framework based on orthogonal multicarriers. This framework supports versatile applications of OFDM and its derivative waveforms within a generalized ISAC system, marking a significant leap in integrating communication and sensing capabilities. A distinguishing feature of this framework is its adaptability, allowing users to intelligently select modulation strategies based on their specific environmental needs. This adaptability optimizes performance across diverse scenarios. Central to our innovations is the proposal of discrete Fourier transform-spread OFDM with index modulation (DFT-S-OFDM-IM). This framework is paired with newly proposed signal processing methods for single-input single-output and multiple-input multiple-output (MIMO) systems. Extensive evaluations highlight DFT-S-OFDM-IM’s superiority, including dramatically reduced peak-to-average power ratios (PAPRs), competitive communication performance, and exceptional sensing capabilities, striking an elegant balance between communication capacity and environmental sensing precision.

The rapid expansion of lithium‐ion batteries (LIBs) is driving the depletion of critical resources and the accumulation of spent batteries, underscoring the urgent need for sustainable recycling strategies. In this study, an innovative recovery strategy that extracts lithium ions from spent battery cathodes, producing high‐purity lithium metal films, is presented. This process eliminates the need for harsh chemicals, such as acids or alkalis, and eliminates wastewater and gas emissions. Electrochemical measurements and characterization demonstrate the successful deintercalation of lithium from multiple spent cathodes, demonstrating the versatility of this approach. Additionally, the synergistic effects of Li‐ion solvation environments and interfacial engineering on enhancing lithium deposition selectivity, achieving a lithium purity of 99.23 wt.%, are emphasized. Moreover, the nucleation and growth mechanisms of lithium metal in the electrodeposition process are investigated, revealing randomly oriented crystallites without preferred orientation along the (110) crystallographic plane. The electrodeposited glassy lithium metal forms a semi‐spherical shape morphology with discontinuous growth, suggesting a sudden influx or accumulation of lithium atoms before the growth direction shifted. Finally, A techno‐economic analysis reveals significant cost advantages of this strategy compared to present recycling methods, while substantially reducing the environmental footprint of spent LIBs processing.

In this paper, we propose an adaptive sieving (AS) strategy for solving general sparse machine learning models by effectively exploring the intrinsic sparsity of the solutions, wherein only a sequence of reduced problems with much smaller sizes need to be solved. We further apply the proposed AS strategy to generate solution paths for large-scale sparse optimization problems efficiently. We establish the theoretical guarantees for the proposed AS strategy including its finite termination property. Extensive numerical experiments are presented in this paper to demonstrate the effectiveness and flexibility of the AS strategy to solve large-scale machine learning models.

As a personality trait, openness to experience plays a central role in driving behaviors associated with exploration and innovation. Despite decades of extensive research, the expression of openness to experience in business strategy remains understudied. The current study addresses this gap by revealing that the openness in experience perceived from American and Chinese CEOs’ social media predicted their firm’s initiation of diversifying M&As, even after controlling for an extensive set of potential confounds. By replicating our findings across two culturally distinct samples, we highlight the role of openness in corporate strategic decision-making, and the usefulness of social media as a source for personality assessment.

Maple seeds are a type of winged seeds that exhibit a unique spinning moving pattern during their descent. We observed in our experiment that the descent trajectories of maple seeds are not uniform, with hypotrochoid and circle being two common regular moving patterns. To explore the underlying physical mechanisms, we constructed a low-speed, vertical, open-type wind tunnel combined with motion capture cameras. A set of simplified six-degree-of-freedom equations for the maple seed model (MSM) was derived based on blade element theory. Building upon this theoretical framework, we implemented a novel numerical method within Simulink to simulate the complete descent process of the maple seed model. Both simulation and experimental results indicate that mass distribution is the dominant factor influencing the flight patterns of maple seeds. The experiments show that when the angle between the third principal inertia axis and the wing span axis is less than 3.25°, the maple seed model rotates around a fixed axis, resulting in a circular trajectory from a top-down view. When the angle exceeds 3.25°, the moving pattern transitions to a hypotrochoid. This finding can be applied to the structural design and flight control of biomimetic aircraft.