Author

Steven W. Su

63 papers · 1999–2026

1999peak 6 in 20122026

Frequent co-authors

Papers

showing 1–25 of 63 RSS

Misperceiving negotiation counterparts: When situationally determined bargaining behaviors are attributed to personality traits.

Michael W. Morris, Richard P. Larrick, Steven K. Su

Sociology and Political Science · Journal of Personality and Social Psychology · July 1999 · cited 141× · DOI ↗ · more like this ·

Several experiments provided evidence that negotiators make systematic errors in personality-trait attributions for the bargaining behaviors of their counterparts. Although basic negotiation behavior is highly determined by bargaining positions, negotiators primarily interpret their counterpart's behavior in terms of the counterpart's personality, such as his or her level of cooperativeness or agreeableness. Data support a model of 4 processes that contribute to misperceptions: (a) the primacy of situations in determining bargaining behavior, (b) the primacy of personality traits in attributions, (c) the lack of sufficient information about the other's situation to discount personality attributions, and (d) the potentially self-confirming consequences of personality attributions for subsequent interactions. The authors discuss implications for research areas such as social cognition in…

iDARTS: Differentiable Architecture Search with Stochastic Implicit Gradients

Miao Zhang, Steven Su, Shirui Pan +3 more

cs.LG · 21 Jun 2021 · cited 91× · arXiv ↗ · more like this ·

\textit{Differentiable ARchiTecture Search} (DARTS) has recently become the mainstream of neural architecture search (NAS) due to its efficiency and simplicity. With a gradient-based bi-level optimization, DARTS alternately optimizes the inner model weights and the outer architecture parameter in a weight-sharing supernet. A key challenge to the scalability and quality of the learned architectures is the need for differentiating through the inner-loop optimisation. While much has been discussed about several potentially fatal factors in DARTS, the architecture gradient, a.k.a. hypergradient, has received less attention. In this paper, we tackle the hypergradient computation in DARTS based on the implicit function theorem, making it only depends on the obtained solution to the inner-loop optimization and agnostic to the optimization path. To further reduce the computational requirements…

Unsupervised machine-learning method for improving the performance of ambulatory fall-detection systems

Mitchell Yuwono, Bruce Moulton, Steven W. Su +2 more

Computer Vision and Pattern Recognition · BioMedical Engineering OnLine · 16 Feb 2012 · cited 91× · DOI ↗ · more like this ·

BACKGROUND: Falls can cause trauma, disability and death among older people. Ambulatory accelerometer devices are currently capable of detecting falls in a controlled environment. However, research suggests that most current approaches can tend to have insufficient sensitivity and specificity in non-laboratory environments, in part because impacts can be experienced as part of ordinary daily living activities. METHOD: We used a waist-worn wireless tri-axial accelerometer combined with digital signal processing, clustering and neural network classifiers. The method includes the application of Discrete Wavelet Transform, Regrouping Particle Swarm Optimization, Gaussian Distribution of Clustered Knowledge and an ensemble of classifiers including a multilayer perceptron and Augmented Radial Basis Function (ARBF) neural networks. RESULTS: Preliminary testing with 8 healthy individuals in a…

Overcoming Multi-Model Forgetting in One-Shot NAS With Diversity Maximization

Miao Zhang, Huiqi Li, Shirui Pan +2 more

Artificial Intelligence · RMIT Research Repository (RMIT University Library) · June 2020 · cited 86× · DOI ↗ · more like this ·

One-Shot Neural Architecture Search (NAS) significantly improves the computational efficiency through weight sharing. However, this approach also introduces multi-model forgetting during the supernet training (architecture search phase), where the performance of previous architectures degrade when sequentially training new architectures with partially-shared weights. To overcome such catastrophic forgetting, the state-of-the-art method assumes that the shared weights are optimal when jointly optimizing a posterior probability. However, this strict assumption is not necessarily held for One-Shot NAS in practice. In this paper, we formulate the supernet training in the One-Shot NAS as a constrained optimization problem of continual learning that the learning of current architecture should not degrade the performance of previous architectures during the supernet training. We propose a…

One-Shot Neural Architecture Search: Maximising Diversity to Overcome Catastrophic Forgetting

Miao Zhang, Huiqi Li, Shirui Pan +4 more

Artificial Intelligence · IEEE Transactions on Pattern Analysis and Machine Intelligence · 3 Nov 2020 · cited 81× · DOI ↗ · more like this ·

One-shot neural architecture search (NAS) has recently become mainstream in the NAS community because it significantly improves computational efficiency through weight sharing. However, the supernet training paradigm in one-shot NAS introduces catastrophic forgetting, where each step of the training can deteriorate the performance of other architectures that contain partially-shared weights with current architecture. To overcome this problem of catastrophic forgetting, we formulate supernet training for one-shot NAS as a constrained continual learning optimization problem such that learning the current architecture does not degrade the validation accuracy of previous architectures. The key to solving this constrained optimization problem is a novelty search based architecture selection (NSAS) loss function that regularizes the supernet training by using a greedy novelty search method to…

High Dimensional Bayesian Optimization via Supervised Dimension Reduction

Miao Zhang, Huiqi Li, Steven Su

cs.LG · 21 Jul 2019 · cited 54× · arXiv ↗ · more like this ·

Bayesian optimization (BO) has been broadly applied to computational expensive problems, but it is still challenging to extend BO to high dimensions. Existing works are usually under strict assumption of an additive or a linear embedding structure for objective functions. This paper directly introduces a supervised dimension reduction method, Sliced Inverse Regression (SIR), to high dimensional Bayesian optimization, which could effectively learn the intrinsic sub-structure of objective function during the optimization. Furthermore, a kernel trick is developed to reduce computational complexity and learn nonlinear subset of the unknowing function when applying SIR to extremely high dimensional BO. We present several computational benefits and derive theoretical regret bounds of our algorithm. Extensive experiments on synthetic examples and two real applications demonstrate the…

Data Clustering Using Variants of Rapid Centroid Estimation

Mitchell Yuwono, Steven W. Su, Bruce Moulton +1 more

Artificial Intelligence · IEEE Transactions on Evolutionary Computation · 11 Sep 2013 · cited 43× · DOI ↗ · more like this ·

Prior work suggests that particle swarm clustering (PSC) can be a powerful tool for solving clustering problems. This paper reviews parts of the PSC algorithm, and shows how and why a new class of algorithms is proposed in an attempt to improve the efficiency and repeatability of PSC. This new implementation is referred to as rapid centroid estimation (RCE). RCE simplifies the update rules of PSC, and greatly reduces computational complexity by enhancing the efficiency of the particle trajectories. On benchmark evaluations with an artificial dataset that has 80 dimensions and a volume of 5000, the RCE variants have iteration times of less than 0.1 s, which compares to iteration times of 2 s for PSC and modified PSC (mPSC). On UC Irvine (UCI) machine learning benchmark datasets, the RCE variants are much faster than PSC and mPSC, and produce clusters with higher purity and greatly…

Self-Organization and Social Organization: U.S. and Chinese Constructions

Steven K. Su, Chi‐yue Chiu, Ying‐yi Hong +3 more

Sociology and Political Science · 4 Apr 2014 · cited 35× · DOI ↗ · more like this ·

From the writings of William James (1890) onward, the construct of the social self has been widely employed in North American and European social psychology (see Markus & Cross, 1990, for a review). Many of the most influential theoretical perspectives in social psychology concern how a per son cognitively represents and emotionally identifies with groups: social identity theory (Hogg & Abrams, 1988; Tajfel & Turner, 1979), social catego rization theory (Turner, 1987), and the group value model (Lind & Tyler, 1988). Although James and many of his Western intellectual heirs have voiced the caveat that the social self is experienced differently in other cultural systems, until very recently there has been little psychological research on this issue.

Robust Neuro-Sliding Mode Multivariable Control Strategy for Powered Wheelchairs

Tuan Nghia Nguyen, Steven W. Su, Hung T. Nguyen

Human-Computer Interaction · IEEE Transactions on Neural Systems and Rehabilitation Engineering · September 2010 · cited 35× · DOI ↗ · more like this ·

This paper proposes an advanced robust multivariable control strategy for a powered wheelchair system. The new control strategy is based on a combination of the systematic triangularization technique and the robust neuro-sliding mode control approach. This strategy effectively copes with parameter uncertainties and external disturbances in real-time in order to achieve robustness and optimal performance of a multivariable system. This novel strategy reduces coupling effects on a multivariable system, eliminates chattering phenomena, and avoids the plant Jacobian calculation problem. Furthermore, the strategy can also achieve fast and global convergence using less computation. The effectiveness of the new multivariable control strategy is verified in real-time implementation on a powered wheelchair system. The obtained results confirm that robustness and desired performance of the…

Estimation of Walking Energy Expenditure by Using Support Vector Regression

Steven W. Su, Lingjiao Wang, Branko G. Celler +2 more

Computer Vision and Pattern Recognition · PubMed · 2005 · cited 32× · DOI ↗ · more like this ·

This paper develops a new predictor of walking energy expenditure from wireless measurements of body movements using triaxial accelerometers. Reliable data were collected from repeated walking experiments in different conditions on a treadmill with simultaneous measurement of expired oxygen and carbon dioxide. Support vector regression, a powerful non-linear regression method, was used to process and model the data. This novel processing method sets this investigation apart from existing papers. Good results were achieved in the robust estimation of walking related energy expenditure from a number of variables derived from triaxial accelerometer and treadmill speed.

Differentiable Neural Architecture Search in Equivalent Space with Exploration Enhancement

Miao Zhang, Huiqi Li, Shirui Pan +3 more

Artificial Intelligence · Monash University Research Portal (Monash University) · 2020 · cited 29× · OpenAlex ↗ · more like this ·

Recent works on One-Shot Neural Architecture Search (NAS) mostly adopt a bilevel optimization scheme to alternatively optimize the supernet weights and architecture parameters after relaxing the discrete search space into a differentiable space. However, the non-negligible incongruence in their relaxation methods is hard to guarantee the differentiable optimization in the continuous space is equivalent to the optimization in the discrete space. Differently, this paper utilizes a variational graph autoencoder to injectively transform the discrete architecture space into an equivalently continuous latent space, to resolve the incongruence. A probabilistic exploration enhancement method is accordingly devised to encourage intelligent exploration during the architecture search in the latent space, to avoid local optimal in architecture search. As the catastrophic forgetting in…

Experimental study on a smart wheelchair system using a combination of stereoscopic and spherical vision

Jordan S. Nguyen, Steven W. Su, Hung T. Nguyen

Human-Computer Interaction · PubMed · July 2013 · cited 28× · DOI ↗ · more like this ·

This paper is concerned with the experimental study performance of a smart wheelchair system named TIM (Thought-controlled Intelligent Machine), which uses a unique camera configuration for vision. Included in this configuration are stereoscopic cameras for 3-Dimensional (3D) depth perception and mapping ahead of the wheelchair, and a spherical camera system for 360-degrees of monocular vision. The camera combination provides obstacle detection and mapping in unknown environments during real-time autonomous navigation of the wheelchair. With the integration of hands-free wheelchair control technology, designed as control methods for people with severe physical disability, the smart wheelchair system can assist the user with automated guidance during navigation. An experimental study on this system was conducted with a total of 10 participants, consisting of 8 able-bodied subjects and 2…

One-Shot Neural Architecture Search via Novelty Driven Sampling

Miao Zhang, Huiqi Li, Shirui Pan +2 more

Artificial Intelligence · Monash University Research Portal (Monash University) · July 2020 · cited 25× · DOI ↗ · more like this ·

One-Shot Neural architecture search (NAS) has received wide attentions due to its computational efficiency. Most state-of-the-art One-Shot NAS methods use the validation accuracy based on inheriting weights from the supernet as the stepping stone to search for the best performing architecture, adopting a bilevel optimization pattern with assuming this validation accuracy approximates to the test accuracy after re-training. However, recent works have found that there is no positive correlation between the above validation accuracy and test accuracy for these One-Shot NAS methods, and this reward based sampling for supernet training also entails the rich-get-richer problem. To handle this deceptive problem, this paper presents a new approach, Efficient Novelty-driven Neural Architecture Search, to sample the most abnormal architecture to train the supernet. Specifically, a single-path…

Shared control strategies for human - Machine interface in an intelligent wheelchair

Anh V. Nguyen, Lien B. Nguyen, Steven W. Su +1 more

Human-Computer Interaction · PubMed · July 2013 · cited 17× · DOI ↗ · more like this ·

In this paper, we introduce a shared control mechanism for an intelligent wheelchair designed to support people with mobility impairments, who also have visual, upper limb, or cognitive impairment. The method is designed to allow users to be involved in the movement as much as possible, while still providing the assistance needed to achieve the goal safely. The data collected through URG-04LX and user interface are analyzed to determine whether the desired action is safe to perform. The system then decides to provide assistance or to allow the user input to control the wheelchair. The experiment results indicate that the method performs effectively with high satisfaction.

Method for increasing the computation speed of an unsupervised learning approach for data clustering

Mitchell Yuwono, Steven W. Su, Bruce Moulton +1 more

Artificial Intelligence · Figshare · June 2012 · cited 17× · DOI ↗ · more like this ·

Clustering can be especially effective where the data is irregular, noisy and/or not differentiable. A major obstacle for many clustering techniques is that they are computationally expensive, hence limited to smaller data volume and dimension. We propose a lightweight swarm clustering solution called Rapid Centroid Estimation (RCE). Based on our experiments, RCE has significantly quickened optimization time of its predecessors, Particle Swarm Clustering (PSC) and Modified Particle Swarm Clustering (mPSC). Our experimental results show that on benchmark datasets, RCE produces generally better clusters compared to PSC, mPSC, K-means and Fuzzy C-means. Compared with K-means and Fuzzy C-means which produces clusters with 62% and 55% purities on average respectively, thyroid dataset has successfully clustered on average 71% purity in 14.3 seconds.

Fast unsupervised learning method for rapid estimation of cluster centroids

Mitchell Yuwono, Steven W. Su, Bruce Moulton +1 more

Artificial Intelligence · UTS ePRESS (University of Technology Sydney) · June 2012 · cited 16× · DOI ↗ · more like this ·

Data clustering is a process where a set of data points is divided into groups of similar points. Recent approaches for data clustering have seen the development of unsupervised learning algorithms based on Particle Swarm Optimization (PSO) techniques. These include Particle Swarm Clustering (PSC) and Modified PSC (mPSC) algorithms for solving clustering problems. However, the PSC and mPSC algorithms tend to be computationally expensive when applied to datasets that have higher levels of dimensionality and large volumes. This paper presents a novel and more efficient swarm clustering strategy we call Rapid Centroid Estimation (RCE). We compare the performance of RCE with the performance of PSC and mPSC in several ways including complexity analyses and particle behavior analyses. Our benchmark testing suggests that RCE can reach a solution 274 times quicker than PSC and 270 times quicker…

Fall detection using a Gaussian distribution of clustered knowledge, augmented radial basis neural-network, and multilayer perceptron

Mitchell Yuwono, Steven W. Su, Bruce Moulton

Artificial Intelligence · UTS ePRESS (University of Technology Sydney) · November 2011 · cited 16× · DOI ↗ · more like this ·

The rapidly increasing population of elderly people has posed a big challenge to research in fall prevention and detection. Substantial amounts of injuries, disabilities, traumas and deaths among elderly people due to falls have been reported worldwide. There is therefore a need for a reliable, simple, and affordable automatic fall detection system. This paper proposes a reliable fall detection algorithm using minimal information from a single waist worn wireless tri-axial accelerometer. The method proposed is to approach fall detection using digital signal processing and neural networks. This method includes the application of Discrete Wavelet Transform (DWT), Regrouping Particle Swarm Optimization (RegPSO), a proposed method called Gaussian Distribution of Clustered Knowledge (GCK), and an Ensemble of Classifiers using two different classifiers: Multilayer Perceptron Neural Network…

The advancement of an obstacle avoidance bayesian neural network for an intelligent wheelchair

Anh V. Nguyen, Lien B. Nguyen, Steven W. Su +1 more

Human-Computer Interaction · PubMed · July 2013 · cited 15× · DOI ↗ · more like this ·

In this paper, an advanced obstacle avoidance system is developed for an intelligent wheelchair designed to support people with mobility impairments who also have visual, upper limb, or cognitive impairment. To avoid obstacles, immediate environment information is continuously updated with range data sampled by an on-board laser range finder URG-04LX. Then, the data is transformed to find the relevant information to the navigating process before being presented to a trained obstacle avoidance neural network which is optimized under the supervision of a Bayesian framework to find its structure and weight values. The experiment results showed that this method allows the wheelchair to avoid collisions while simultaneously navigating through an unknown environment in real-time. More importantly, this new approach significantly enhances the performance of the system to pass narrow openings…

Optimum Experimental Design applied to MEMS accelerometer calibration for 9-parameter auto-calibration model

Lin Ye, Steven W. Su

Computer Networks and Communications · PubMed · August 2015 · cited 14× · DOI ↗ · more like this ·

Optimum Experimental Design (OED) is an information gathering technique used to estimate parameters, which aims to minimize the variance of parameter estimation and prediction. In this paper, we further investigate an OED for MEMS accelerometer calibration of the 9-parameter auto-calibration model. Based on a linearized 9-parameter accelerometer model, we show the proposed OED is both G-optimal and rotatable, which are the desired properties for the calibration of wearable sensors for which only simple calibration devices are available. The experimental design is carried out with a newly developed wearable health monitoring device and desired experimental results have been achieved.

An algorithm for scalable clustering: Ensemble Rapid Centroid Estimation

Mitchell Yuwono, Steven W. Su, B. D. Moulton +2 more

Artificial Intelligence · UTS ePRESS (University of Technology Sydney) · July 2014 · cited 13× · DOI ↗ · more like this ·

This paper describes a new algorithm, called Ensemble Rapid Centroid Estimation (ERCE), designed to handle large-scale non-convex cluster optimization tasks, and estimate the number of clusters with quasi-linear complexity. ERCE stems from a recently developed Rapid Centroid Estimation (RCE) algorithm. RCE was originally developed as a lightweight simplification of the Particle Swarm Clustering (PSC) algorithm. RCE retained the quality of PSC, greatly reduced the computational complexity, and increased the stability. However, RCE has certain limitations with respect to complexity, and is unsuitable for non-convex clusters. The new ERCE algorithm presented here addresses these limitations.

Advanced robust tracking control of a powered wheelchair system

Nghia T. Nguyen, Hung T. Nguyen, Steven W. Su

Human-Computer Interaction · Conference proceedings · August 2007 · cited 13× · DOI ↗ · more like this ·

In this paper, the dynamic multivariable model of the wheelchair system is obtained including the presence of transportation lags. The triangular diagonal dominance (TDD) decoupling technique is applied to reduce this multivariable control problem into two independent scalar control problems. An advanced robust control technique for the wheelchair has been developed based on the combination of a TDD decoupling strategy and neural network controller design. The results obtained from the real-time implementation confirm that robust performance for this multivariable wheelchair control system can indeed be achieved.

EMG Signals Classification on Human Activity Recognition using Machine Learning Algorithm

Ku Nurhanim, Irraivan Elamvazuthi, Lila Iznita Izhar +2 more

Computer Vision and Pattern Recognition · 2021 8th NAFOSTED Conference on Information and Computer Science (NICS) · 21 Dec 2021 · cited 12× · DOI ↗ · more like this ·

In Human activity recognition (HAR) research, it is a common practice to use wearable sensors to acquire the signals for human daily activities. In this study, an experimental data from electromyography (EMG) wireless sensors is analysed for six different activities recognition. This paper aims to compare EMG signals of left and right of upper leg muscles by using Random Forest (RF) Machine Learning Classifier. The HAR processing comprises of data filtering and segmentation, data feature extraction, feature selection of the data, and classification. Model evaluation of holdout method is implemented for classification assessment. The performance of all human daily activities is evaluated according to the comparison of precision and recall for each activity. The results show combined muscles obtained the highest precision and recall on running activity with 89.2% and 88.3%. The highest…

Multirealization of linear systems

Brian D. O. Anderson, Steven W. Su, Thomas S. Brinsmead

Artificial Intelligence · IEEE Transactions on Circuits and Systems II Analog and Digital Signal Processing · August 2005 · cited 11× · DOI ↗ · more like this ·

For multiple-model adaptive control systems, "multi-controller" architecture can be efficiently implemented (multirealized) by means of a "state-shared" parameter-dependent feedback system. Necessary and sufficient conditions for the multirealization of a family of linear multivariable systems based on matrix fractional descriptions are presented. The problem of the minimal generic multirealization of a set of linear systems is introduced and solved.

Optimization strategies for rapid centroid estimation

Mitchell Yuwono, Steven W. Su, B. D. Moulton +1 more

Artificial Intelligence · PubMed · August 2012 · cited 10× · DOI ↗ · more like this ·

Particle swarm algorithm has been extensively utilized as a tool to solve optimization problems. Recently proposed particle swarm±based clustering algorithm called the Rapid Centroid Estimation (RCE) is a lightweight alteration to Particle Swarm Clustering (PSC). The RCE in its standard form is shown to be superior to conventional PSC algorithm. We have observed some limitations in RCE including the possibility to stagnate at a local minimum combination and the restriction in swarm size. We propose strategies to optimize RCE further by introducing RCE+ and swarm RCE+. Five benchmark datasets from UCI machine learning database are used to test the performance of these new strategies. In Glass dataset swarm RCE+ is able to achieve highest purity centroid combinations with less iteration (90.3%±1.1% in 9±5 iterations) followed by RCE+ (89%±3.5% in 65±62 iterations) and RCE (87%±5.9% in…

Universal algorithm for exercise rate estimation in walking, cycling and rowing using triaxial accelerometry

Tingli Cheng, A.V. Savkin, Branko G. Celler +2 more

Signal Processing · Electronics Letters · 8 Apr 2009 · cited 10× · DOI ↗ · more like this ·

A technique that can reliably monitor exercise intensity plays an important role for the effectiveness and safety of an exercise prescription. A universal algorithm for the recursive estimation of exercise rate during a variety of aerobic exercises using measurements from a body-mounted triaxial accelerometer (TA) is proposed. Information about the type of exercise is not required by the algorithm and the TA can be mounted at the same location regardless of the exercise type. The algorithm involves period detection and data fusion. Experimental results demonstrate that the algorithm is effective for common aerobic exercises.