Singapore University of Technology and Design
Recent publications
Joint network topology inference represents a canonical problem of jointly learning multiple graph Laplacian matrices from heterogeneous graph signals. In such a problem, a widely employed assumption is that of a simple common component shared among multiple graphs. However, in practice, a more intricate topological pattern, comprising simultaneously of homogeneous and heterogeneous components, would exhibit in multiple graphs. In this paper, we propose a general graph estimator based on a novel structural fusion regularization that enables us to jointly learn multiple graphs with such complex topological patterns, and enjoys rigorous theoretical guarantees. Specifically, in the proposed regularization term, the structural similarity among graphs is characterized by a Gram matrix, which enables us to flexibly model different types of network structural similarities through different Gram matrix choices. Algorithmically, the regularization term, coupling the parameters together, makes the formulated optimization problem intractable, and thus, we develop an implementable algorithm based on the alternating direction method of multipliers (ADMM) to solve it. Theoretically, non-asymptotic statistical analysis is provided, which precisely characterizes the minimum sample size required for the consistency of the graph estimator. This analysis also provides high-probability bounds on the estimation error as a function of graph structural similarities and other key problem parameters. Finally, the superior performance of the proposed method is demonstrated through simulated and real data examples.
In this paper, we construct q -ary two-deletion correcting codes and burst-deletion correcting codes, where q ≥ 2 is an even integer. For two-deletion codes, our construction has redundancy 5 log n + O (log q log log n ) and has encoding complexity near-linear in n , where n is the length of the message sequences. For burst-deletion codes, we first present a construction of binary codes with redundancy log n +9 log log n + γ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</sub> + o (log log n ) bits (γ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</sub> is a constant that depends only on t ) and capable of correcting a burst of at most t deletions, which improves the Lenz-Polyanskii Construction (ISIT 2020). Then we give a construction of q -ary codes with redundancy log n +(8 log q + 9) log log n + γ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</sub> + o (log log n ) bits and capable of correcting a burst of at most t deletions.
The impact of intracell magnetic coupling on spin transfer torque-magnetic random access memory (STT-MRAM) device-to-device variability has been investigated based on experiments and simulations. Measured switching voltages of CoFeB/MgO STT-MRAMs are found to be directly correlated with offset fields extracted from magnetic measurements of the same set of devices. By spintronic and Monte Carlo simulations incorporating stochasticity and variability of the STT-MRAMs, the origin of the experimentally observed correlation has been traced back to the stray field arising from intracell magnetic coupling. Based on this finding, READ and write error rates (WERs) of STT-MRAMs have been evaluated in the presence of a stray field, employing the Fokker–Planck (FP) equation-based numerical model, calibrated against switching probabilities of measured devices. The results of the statistical simulation show that depending on the direction and magnitude of the stray field, READ and WERs preferentially increase in one switching direction, while decreasing in the opposite direction. The pulse widths required to bring the WRITE and READ error rates to a nominal value of $\text{10}^{-\text{6}}$ have been estimated considering $\pm$ 20% variation of the stray field. It is also shown that READ operations of larger devices are more susceptible to intracell magnetic coupling-induced variability than the WRITE operations.
  • Rebecca Erbanni
    Rebecca Erbanni
  • Xiansong Xu
    Xiansong Xu
  • Tommaso F. Demarie
    Tommaso F. Demarie
  • Dario Poletti
    Dario Poletti
Digital quantum computers have the potential to study the dynamics of many-body quantum systems. Nonequilibrium open quantum systems are, however, less straightforward to be implemented. Here we explore the feasibility of studying steady-state transport in strongly interacting many-body quantum systems on a digital quantum computer. To do so, we consider a collisional model representation of the nonequilibrium open dynamics for a boundary-driven XXZ spin chain. More specifically, we investigate how the depth of the quantum circuit is affected by how close we want the steady state to be to the one expected from the underlying master equation. We study the simulation of a boundary-driven spin chain in regimes of weak and strong interactions, which would lead in large systems to diffusive and ballistic dynamics, considering also possible errors in the implementation of the protocol. Last, we analyze the effectiveness of digital simulation via the collisional model of current rectification when the XXZ spin chains are subject to nonuniform magnetic fields and show that, although the circuit depths required to reach steady states are still prohibitive for today's hardware, few collisions are enough to suggest a strong rectifying power.
Transactional stream processing (TSP) strives to create a cohesive model that merges the advantages of both transactional and stream-oriented guarantees. Over the past decade, numerous endeavors have contributed to the evolution of TSP solutions, uncovering similarities and distinctions among them. Despite these advances, a universally accepted standard approach for integrating transactional functionality with stream processing remains to be established. Existing TSP solutions predominantly concentrate on specific application characteristics and involve complex design trade-offs. This survey intends to introduce TSP and present our perspective on its future progression. Our primary goals are twofold: to provide insights into the diverse TSP requirements and methodologies, and to inspire the design and development of groundbreaking TSP systems.
Fabricating a functional heterogeneous interface to enhance catalytic performance is quite significant for developing high-efficiency electrocatalysts. Herein, a coral-like nickel phosphide@cerium oxide (Ni2P@CeO2) hybrid nanoarray on nickel foam was designed via selective-phosphorization of nickel hydroxide@cerium oxide (Ni(OH)2@CeO2). Benefiting from CeO2 as the “electron pump,” it leads to electron transfer from Ni2P to the CeO2 side, and induces electron redistribution in the interface boundary, thereby optimizing the H* adsorption free energy in the hydrogen evolution reaction (HER) process. As hypothesized, the water molecules will preferentially adsorb on the CeO2 side due to its better affinity for oxygen-containing species, and will readily break down into OH* and H* at a lower energy barrier. Subsequently, benefiting from the lower H* adsorption free energy of P sites, the generated H* will migrate to the Ni2P side through the spillover process. Contributing to the synergistic effect of double-active sites, the Ni2P@CeO2/NF electrode exhibits brilliant catalytic performance for HER with 62 mV to attain 10 mA/cm2 and exceptional durability over 100 h in alkaline solution at ~ 100 mA/cm2. Meanwhile, attributing to the similar interface electron redistribution effect, the precursor Ni(OH)2@CeO2/NF likewise displays excellent oxygen evolution reaction (OER) electrocatalytic performance, which only requires 229 mV to arrive at 10 mA/cm2, even better than benchmark ruthenium dioxide (RuO2). Hence, the assembled Ni(OH)2@CeO2/NF||Ni2P@CeO2/NF system only needs 1.53 V to achieve 10 mA/cm2 in alkaline solution. Moreover, the electrolyzer also presents brilliant electrocatalytic activity and stability in alkaline natural seawater electrolyte with higher reserves on earth. “Electrons pump” effect of CeO2 ensures that interface-engineered Ni2P@CeO2 hybrid nanoarrays prepared via selective-phosphorization treatment present superior HER catalytic performance
Communicating with others while engaging in simple daily activities is both common and natural for people. However, due to the hands- and eyes-busy nature of existing digital messaging applications, it is challenging to message someone while performing simple daily activities. We present GlassMessaging, a messaging application on Optical See-Through Head-Mounted Displays (OHMDs), to support messaging with voice and manual inputs in hands- and eyes-busy scenarios. GlassMessaging is iteratively developed through a formative study identifying current messaging behaviors and challenges in common multitasking with messaging scenarios. We then evaluated this application against the mobile phone platform on varying texting complexities in eating and walking scenarios. Our results showed that, compared to phone-based messaging, GlassMessaging increased messaging opportunities during multitasking due to its hands-free, wearable nature, and multimodal input capabilities. The affordance of GlassMessaging also allows users easier access to voice input than the phone, which thus reduces the response time by 33.1% and increases the texting speed by 40.3%, with a cost in texting accuracy of 2.5%, particularly when the texting complexity increases. Lastly, we discuss trade-offs and insights to lay a foundation for future OHMD-based messaging applications.
Two-photon polymerization lithography is promising for producing three-dimensional structures with user-defined micro- and nanoscale features. Additionally, shrinkage by thermolysis can readily shorten the lattice constant of three-dimensional photonic crystals and enhance their resolution and mechanical properties; however, this technique suffers from non-uniform shrinkage owing to substrate pinning during heating. Here, we develop a simple method using poly(vinyl alcohol)-assisted uniform shrinking of three-dimensional printed structures. Microscopic three-dimensional printed objects are picked and placed onto a receiving substrate, followed by heating to induce shrinkage. We show the successful uniform heat-shrinking of three-dimensional prints with various shapes and sizes, without sacrificial support structures, and observe that the surface properties of the receiving substrate are important factors for uniform shrinking. Moreover, we print a three-dimensional mascot model that is then uniformly shrunk, producing vivid colors from colorless woodpile photonic crystals. The proposed method has significant potential for application in mechanics, optics, and photonics.
As a symbol-level precoding scheme, constructive interference precoding (CIP) has been demonstrated its superiority in multi-antenna orthogonal multiple access (OMA). By utilizing both the channel state information (CSI) and data symbols, harmful multiuser interference can be converted into useful reception power via the well-designed CIP. When CIP meets non-orthogonal multiple access (NOMA) whose bottleneck is usually at the weaker user, this paper is the first to propose CIP to enhance the downlink MISO-NOMA networks, by making the desired signal of the stronger user in a typical NOMA pair constructive to the weaker user. In our CIP-NOMA scheme, we properly design the CIP precoder for transmit power minimization at the base station (BS), subject to signal-to-interference-plus-noise ratio (SINR) requirements of NOMA users. We further derive its closed-form solutions with Karush-Kuhn-Tucker (KKT) conditions, and optimally obtain the desired CIP precoders. Moreover, as compared to conventional NOMA schemes, we theoretically prove that once two NOMA users possess distinct channel gains, our optimized CIP-NOMA scheme always uses lower transmit power to reach the SINR thresholds. To be robust against the channel estimation errors, we extend our CIP-NOMA scheme to the scenario of imperfect CSI, by further addressing the hidden CSI errors. Specifically, we first introduce some auxiliary variables to separate the coupled vectors, and then use S-Procedure and semi-definite relaxation (SDR) to further transform them into convex ones. Extensive simulations verify that our CIP-NOMA scheme greatly outperforms the benchmarks with both perfect and imperfect CSI. Index Terms-Constructive interference precoding, NOMA, semi-definite programming, imperfect CSI.
When a complex cyber-physical infrastructure is attacked, operators need to isolate the attack location. Since sensors and actuators are physically intertwined in such structures, operators must be able to separate incoming status data to isolate the precise location of the cyberattack. We let several unsupervised algorithms compete and analyze the extent to which they can provide fast and efficient analysis in order to support operators with this task, using data from the Secure Water Treatment testbed (SWaT), an experimental infrastructure in Singapore that allows us to simulate the behavior of large infrastructure systems. We find that the k-Shape algorithm performs best. This result suggests that unsupervised algorithms can support human operators efficiently even in critical infrastructures with complex sensor data time series.
We recommend three architectural choices which should contribute to a more effective defense of cyber-physical systems in general and industrial control systems in particular: mixed reality solutions for a more effective control and response, a zero trust architecture that mitigates human fallacies, and automated defense systems based on a security by design approach. We illustrate and discuss the implications of these choices.
The formation of a MgCO3 shell hampers CO2 capture efficiency in MgO. Our previous studies developed MgO/Mg(OH)2 composites to facilitate CO2 diffusion, improving capture efficiency. However, MgCO3 still formed along the interfaces. To tackle this issue, we engineered the MgO/Mg(OH)2 interfaces by incorporating Cl⁻, SO4²⁻, and PO4³⁻ additives. Novel MgO–H2O–MgX (X = Cl⁻, SO4²⁻, and PO4³⁻) composites were synthesized to explore the role of additives in preventing MgCO3 formation. MgO–Mg(OH)2–MgCl2 nano-composites displayed enhanced CO2 adsorption and stability. This breakthrough paves the way for effective bio-inspired strategies in overcoming CO2 transport barriers in MgO-based adsorbents.
Floor-cleaning robots are primarily designed to clean on a single floor, while multi-floor environments are usually not considered target applications. However, it is more efficient to have an autonomous floor-cleaning robot that can climb stairs and reach the next floors in a multi-floor building. To operate in such environments, the ability of a mobile robot to autonomously traverse staircases is very important. For this operation, staircase detection and localization are essential components for planning the traversal route on staircases. This article describes a deep learning approach using a convolutional neural network (CNN)-based robot operation system (ROS) framework for staircase detection, localization, and maneuvering of the robot to the detected stair. We present a real-time object detection framework to detect staircases in incoming images. We also localize these staircases using a contour detection algorithm to detect the target point: a point close to the center of the first step, and an angle of approach to the target point with respect to the current location of the robot. Experiments are performed with data from images captured on different types of staircases at different viewpoints/angles. The experimental results show that the presented approach can achieve an accuracy of 95% and a recall of 86.81%. A total runtime of 155 ms is taken to identify the presence of a staircase and the detection of the first step in the working environment, as well as being able to locate the target point with an accuracy of ±2 cm, ±1 degree.
In this research, the impact of pressurization (300–600 MPa) on the oscillatory rheology of a blend of xanthan gum (XG) and guar gum (GG) in solution (XG/GG) was examined at a selected gum concentration using a response surface methodology. Three independent variables pressure (0.101–600 MPa) concentration (0.75%–1.25%), and temperature (40–70°C)—were employed to maximize the dynamic moduli (G′, G″, and η*) of the blend solutions. The developed model suggested that all the linear, two quadratics (concentration and temperature), and two interactions (concentration‐pressure and concentration‐temperature) terms were significant (p < 0.05) for the dynamic moduli (R² = 0.99) with insignificant lack‐of‐fit. The pressure‐induced mechanical rigidity was found to be higher when compared with the heat‐treated solutions, which requires further comprehensive structural assessment to endorse the order–disorder transition produced by high pressure. The superposition principles correlating time–temperature and time–pressure adequately fit the pressurized blend solutions. Practical applications The influence of high‐pressure on a blend of gum containing xanthan and guar has industrial significance by improving the structure/texture, and consistency of food products. The synergic action of gums enables them to act as a fat replacer and texture modifier in HP‐treated foods for special needs, in particular dysphagia diets.
Reconfigurable robots have the potential to perform complex tasks by adapting their morphology to different environments. However, designing optimal morphologies for these robots is challenging due to the large design space and the complex interactions between the robot and the environment. An in-house robot named Smorphi, having four holonomic mobile units connected with three hinge joints, is designed to maximize area coverage with its shape-changing features using transformation design principles (TDP). The reinforcement learning (RL) approach is used to identify the optimal morphologies out of a vast combination of hinge angles for a given task by maximizing a reward signal that reflects the robot's performance. The proposed approach involves three steps: (i) Modeling the Smorphi design space with a Markov decision process (MDP) for sequential decision-making; (ii) a footprint-based complete coverage path planner to compute coverage and path length metrics for various Smorphi morphologies; and (iii) pptimizing policies through proximal policy optimization (PPO) and asynchronous advantage actor-critic (A3C) reinforcement learning techniques, resulting in the generation of energy-efficient, optimal Smorphi robot configurations by maximizing rewards. The proposed approach is applied and validated using two different environment maps, and the results are also compared with the suboptimal random shapes along with the Pareto front solutions using NSGA-II. The study contributes to the field of reconfigurable robots by providing a systematic approach for generating optimal morphologies that can improve the performance of reconfigurable robots in a variety of tasks.
Aqueous rechargeable Zn metal batteries (ARZBs) are extensively studied recently because of their low-cost, high-safety, long lifespan, and other unique merits. However, the terrible ion conductivity and insufficient interfacial redox dynamics at low temperatures restrict their extended applications under harsh environments such as polar inspections, deep sea exploration, and daily use in cold regions. Electrolyte modulation is considered to be an effective way to achieve low-temperature operation for ARZBs. In this review, first, the fundamentals of the liquid-solid transition of water at low temperatures are revealed, and an in-depth understanding of the critical factors for inferior performance at low temperatures is given. Furthermore, the electrolyte modulation strategies are categorized into anion/concentration regulation, organic co-solvent/additive introduction, anti-freezing hydrogels construction, and eutectic mixture design strategies, and emphasize the recent progress of these strategies in low-temperature Zn batteries. Finally, promising design principles for better electrolytes are recommended and future research directions about high-performance ARZBs at low temperatures are provided.
This paper proposes a novel method that addresses a non-traditional class of outlier detection problems. The purpose of most outlier detection methods in the literature is to detect outliers within a dataset. A record can be considered as an outlier if it is distinct from the regular records in the dataset. However, the purpose of the novel outlier detection method proposed by this paper is to detect outlier data groups (a data group may denote a site or a project) with respect to a soil/rock property database. A data group is an outlier group if its characteristics (mean, variance, correlation, or higher order dependency) are distinct from the regular data groups in the database. This paper frames the outlier detection problem into a formal hypothesis testing problem with the null hypothesis “the target data group is identically distributed as the regular groups in the database”. With the hierarchical Bayesian model (HBM) previously developed by the first two authors, the p-value for this hypothesis testing problem can be estimated rigorously. Numerical and real examples show that the p-value can effectively detect outlier data groups as well as outlier records with respect to a database.
Pair-wise co-mutation networks of the mitochondrial genome have already provided ample evidences about the roles of genetic interactions in the manifestation of phenotype under altered environmental conditions. Here, we present a method to construct and analyze higher-order interactions, namely, 3-uniform hypergraphs of the mitochondrial genome for different altitude populations to decipher the role of co-mutating variable sites beyond pair-wise interactions. While the weights distribution of such gene hyperedges manifested power-law for all the altitudes, we identified altitude-specific genes based on gene hyperedge weight. This framework of hypergraphs serves a promising avenue for future investigation of nuclear genomes in context of phenotypic association and genetic disorders.
Institution pages aggregate content on ResearchGate related to an institution. The members listed on this page have self-identified as being affiliated with this institution. Publications listed on this page were identified by our algorithms as relating to this institution. This page was not created or approved by the institution. If you represent an institution and have questions about these pages or wish to report inaccurate content, you can contact us here.
1,699 members
Alexander Binder
  • Information Systems Technology and Design Pillar
Balakrishnan Ramalingam
  • Division of Engineering Product Development
Joel K.W. Yang
  • Division of Engineering Product Development
Arlindo Silva
  • Engineering Product Development Pillar
Hoang Son Dau
  • SUTD-MIT International Design Centre
Information
Address
8 Somapah Road, 487372, Singapore, Singapore
Website
http://www.sutd.edu.sg/