Feature Selection and Occupancy Classification Using Seismic Sensors

Size: px
Start display at page:

Download "Feature Selection and Occupancy Classification Using Seismic Sensors"

Transcription

1 Feature Selection and Occupancy Classification Using Seismic Sensors Arun Subramanian 1, Kishan G. Mehrotra 1, Chilukuri K. Mohan 1, Pramod K. Varshney 1 and Thyagaraju Damarla 2 1 Department of Electrical Engineering and Computer Science, Syracuse University, Syracuse, NY 13244, USA 2 Army Research Laboratory, Adelphi, MD 2783, USA Abstract. In this paper, we consider the problem of indoor surveillance and propose a feature selection scheme for occupancy classification in an indoor environment. The classifier aims to determine whether there is exactly one occupant or more than one occupant. Data are obtained from six seismic sensors (geophones) that are deployed in a typical building hallway. Four proposed features exploit amplitude and temporal characteristics of the seismic time series. A neural network classifier achieves performance ranging between 77% to 95% on the test data, depending on the type of construction of the location in the building being monitored. 1 Introduction Automatic surveillance is essential for many scenarios that require remote monitoring. Several sensors of different modalities (e.g., video, acoustic, infra-red and seismic) may be deployed to monitor areas such as international borders or cleared buildings where a sustained presence of personnel is not feasible. In some cases, constraints on logistics do not permit the use of information rich sensors such as video. We consider one such scenario for indoor surveillance and investigate the use of seismic sensors for occupancy classification. Detecting the presence of objects using seismic sensors can be categorized into two broad areas: detection of (i) humans, and (ii) other objects (such as vehicles). Detection of vehicles is easier to accomplish than detection of humans, since the seismic signal from a vehicle has a higher signal to noise ratio (see [1 3]). The harmonic signatures of two vehicles of the same type are consistent, and those of different vehicle types are generally distinguishable. In contrast, classification of human occupancy using footstep signals is difficult in that different people walk with different gaits and at varying pace, resulting in vibration patterns that are difficult to identify uniquely. To detect occupants based on footstep signal processing using seismic sensors, researchers use approaches such as auto-regressive modeling, signal moments, This research was sponsored by Army Research Laboratory and was accomplished under Cooperative Agreement No. W911NF It was also supported in part by ARO grant W911NF

2 time-scale analysis and explicit experimental modeling [4 7]. Using a copula based approach, Iyengar et al. [8] fuse signals from acoustic and seismic sensors for footstep detection. Although the detection of the presence of humans has been successfully addressed by such researchers, finer-grain analysis is much more difficult. One such challenging task, addressed in this paper, is the determination of whether one or more people are walking together in the environment being monitored. This is because people walking together have a psychological tendency to walk in lock step [9]. Since the raw signals are not easily classifiable, an important step is to derive suitable features that can assist classification. One of the main contributions of this work is to develop features that enable us to distinguish between the presence of one or more than one person in the given region of interest. In Section 2, we describe the data collection process. The raw data is preprocessed to extract useful information, and to eliminate noise, using a methodology based on empirical mode decomposition (EMD), described in Section 3. Our approach for feature selection is presented in Section 4. Section 5 contains the classification results and Section 6 contains our conclusions. 2 Data collection Six GS 2DX geophones were used for the purpose of data collection. The geophones are designed to be floor mounted. The sensors were configured as a linear array. Data was collected in two (different) building hallways of similar construction. The sensors were placed along the long edge of the hallway. The distance between two adjacent sensors was maintained at 5ft. Data was acquired using a 16 bit A/D converter at a sampling rate of 5kHz. The raw signal was uniformly down-sampled to 124 per second. The approximate duration of the data collected per trial is 12 seconds. Multiple persons participated in the data collection. The footstep data thus collected consists of 12 single-person trials (i.e., a given trial has exactly one participant walking along the hallway) and 12 two-person trials (a given trial has exactly two participants walking along the hallway). Each dataset consists of 6 trials from Building 1 and 6 trials from Building 2. 3 Overview of the Preprocessing In order to derive reliable features from pseudo-periodic signals like footsteps, it is of interest to extract a clean envelope corresponding to footfalls. We have used empirical mode decomposition (EMD) [1] for this purpose. EMD is a data-driven decomposition technique that captures oscillations at several scales. Each of these scales is called a mode and the function corresponding to a mode is called an intrinsic mode function (IMF). In order to capture interesting characteristics of footsteps and eliminate inherent noise present in the data, we have used only a small subset of these modes based on the total variation (TV) norm [6] such that the sum of TV norm of the selected modes is 9% of the total

3 TV norm. For a time-series of length T, these modes are treated as random vectors with T observations. These vectors are linearly combined, using the first principal component. 1 The raw data is denoted by y (k) (t) and the processed data so obtained is denoted by x (k) (t) for i = 1,..., 6, j = 1,..., 12, k = 1, 2, t = 1,..., T ; where i represents the sensor, j refers to trial, k refers to number of persons, and t is the time index. This processed data is used in the rest of the paper to extract features for classification. 2 The effect of this preprocessing is seen in1 Fig. 3. The EMD based processing extracts a faithful envelope of the raw signal revealing the salient features of -1 each footfall (a) (b) 1.5 Fig. 11. A comparison of a two-second segment of the raw signal y (2) of envelope.5 extraction, x (2) (t), for the 2 persons case Feature Selection (t) and the result In this section, we present four features that were employed to achieve classification between presence of one person versus two persons. We assume that the detection has already been accomplished successfully, i.e., it is certain that at least one occupant is present. The seismic signal of a sensor decays rather rapidly as person(s) move away from the sensor. For each sensor separately, we, therefore, extract a five second segment of the useful data as proposed below; the discarded data is deemed to be non-informative. For each i, j, k, first we find the time index where x (k) is maximum and extract data for all time indices that are contained in ±t o on either side of the index where the maximum occurs. We choose t = , because, as stated earlier, we collect 124 samples per second. For instance, if t max was the time when x (k) takes it s maximum value, then the extracted data will be {x (k) (t) : t = t max t, t max t + 1,..., t max + t }. However, for notational convenience we renumber the time indices and denote the extracted data as {z (k) (t) : t =, 1,..., 2t }, 1 We have observed that about 8% of the signal energy is contained in the first principal component.

4 person 2 person.6 Amplitude Frequency (Hz) Fig. 2. Periodogram comparison of 1 person and 2 person trials. for all values of (i, j, k). Note that this data will be extracted from the early part of {x (k) (t)} for the first sensor (i = 1) and later part for the sixth sensor (i = 6). In the following subsections, features are extracted from these extracted data sets only. 4.1 Periodogram and autocorrelation Footsteps are quasi-periodic signals. Typically, the periodogram and autocorrelation function (ACF) are used in determining signal periodicity [2, 3, 11]. But, analysis by Houston and McGaffigan [9] shows that the use of spectral measures is not useful for counting the number of personnel in the region of interest. In the following discussion, we note that while the periodogram and autocorrelation are not directly useful for our occupancy classification problem, nevertheless they provide some valuable insights. Fig. 2 shows a comparison of the periodograms of 2 trials of the one person and two person cases. The figure suggests that the periodogram alone may be insufficient for the classification task at hand. We expect that heel-toe transitions will be borne out in the single occupant case and will be blurred in the case of two or more occupants. In order to find support for this argument, we plotted the mean of the ACF across sensors for a single period (Fig. 4.1). We use the peak frequency from the periodogram to obtain the period of the ACF. Clearly, the ACF for one person goes through a typical pattern of local maxima and minima as heel and toe spikes align with each other. When the ACF for the first period are averaged over all the training data for all sensors in the one-person case, we obtain the ACF template seen in Fig. 3(a). Therefore, the feature extraction procedure is, 1. Form the template (Fig. 3(a)) 2. Calculate the area which measures the difference between the template and the ACF for a given trial (see Fig. 4). This difference is the mean-square

5 (a) (b) Fig. 3. The autocorrelation function for a single period averaged over all sensors. (a) 1 person. (b) 2 persons. X-axis τ is in seconds. Y -axis is r (τ). Fig. 4. Comparison of autocorrelation functions. The shaded region shows the difference in area which, when integrated, gives the mean-square error between the template and the ACF for the given trial. X-axis: τ, Y -axis: r (τ). error between the two curves and is denoted as MSE (k), which is the first of the four features that we use for occupancy classification. 4.2 Signal energy Variance is a measure of signal energy and serves as an intuitive indicator of occupancy. That is, when the number of occupants is large, the signal energy will be large. Using the extracted data described earlier we calculate the variance as described below. [ S (k) ] 2 = 1 2t o 2t o ( t= z (k) (t) z(k) ) 2, (1)

6 where z (k) is the sample mean (of the 2t observations). These variances are averaged by taking their mean across all six sensors, and S (k) j = 1 6 N [ i=1 S (k) is used as one of the features to be used for classification. ] 2 (2) 4.3 Ratio of time spent in states We expect that footsteps produce a quasi-periodic time series when one person walks whereas footsteps for two persons will be irregular and spread out. For two or more persons this footstep state will occupy a greater proportion of a window of fixed duration. Based on this criterion, the feature R(states) is calculated for each value of (i, j, k) as described below. For notational convenience we do not use j and k below. As stated earlier, we use the extracted data only. We move a sliding window (one-tenth of a second long) over this data, calculate the average of the x-values within the window, and compare it with a threshold η to calculate { 1 if the average of the observations within the window is > η d pi = if the average of the observations within the window is η (3) where p = 1,..., P and P represents the number of times the sliding window fits over the data. In the ideal situation, each 1 represents the Footstep state and each represents the Silence state. The ratio p,i R(states) = d pi p,i (1 d (4) pi) captures the ratio of time spent in these states versus not in the state. These ratios, the third feature, are obtained for all values of (j, k) and are denoted as R(states) (k) j. Threshold η is calculated from the first.25 seconds of the x data which represents the background process. Let µ b and σ b denote the sample mean and standard deviation calculated from this initial.25 second period. The threshold is then set as, η = µ b + 3σ b (5) 4.4 Cross-correlation and wavelet based feature The cross-correlation sequence between adjacent sensors is one way to measure the similarity between the data collected by two sensors. We considered the crosscorrelation between the two nearest neighbors of a given sensor; for example, two nearest neighbors of sensor 2 are sensor 1 and sensor 3, respectively. As before the

7 extracted data is obtained for sensors i = 2, 3, 4, 5 and the following correlation functions are computed: r i+ (τ) = Corr(z (k) (t), z(k) (i+1),j(t + τ)) (6) r i (τ) = Corr(z (k) (t), z(k) (i 1),j(t + τ)) (7) r i± (τ) = Corr(z (k) (i+1),j(t), z(k) (i 1),j(t + τ)) (8) R i (τ) = r i+ (τ)r i (τ) r i± (τ) (9) where τ extends from -124 to 124 (i.e., up to a shift of 1 second in both directions). We expect that, in the ideal situation, R i = for the 1-person case and R i when two or more persons are walking. In practice, R i is expected to take a small value for the 1-person case and a large value in the 2 or more persons case. In Fig. 5, we observe wide major peaks and thinner auxiliary peaks in the two-person case. In other words, the 2-person information is contained at different scales. This suggests that wavelet decomposition would be useful in extracting useful information from R i (τ) person 2 persons Fig. 5. Typical plot of R i vs. τ We chose the Mexican hat wavelet and calculated the continuous wavelet transform (scalogram) of {R i (τ) : τ = 124 to 124}, C i (a, b) = R i (τ)ψ (τ b, a)dτ, (1) allτ 1 ψ (τ, a) = (1 τ 2 ) 2πa 3 a 2 exp ( τ 2 ) 2a 2

8 We note that, in theory, the scale a can have an infinite number of values. For numerical computation, we chose the maximum value of a to be 1/(2f ) where f is the peak frequency from the periodogram of the extracted data. Finally, a summary statistics can be measured in terms of the average S C, C(a, b) = 1 4 S C = b 5 C i (a, b) (11) i=2 or use this feature separately for each sensor as: max C(a, b) (12) a S Ci = b max C i (a, b). (13) a 5 Classification Procedure and Results Section 4 discussed the features investigated for classification problem of determining one vs. two occupants in a building hallway. There are two possible fusion schemes for the selected features: 1. Use the 4-dimensional feature vector F = [MSE (k) j, S (k) j, R(states) (k) j, S C ] where the individual features have been fused across sensors by taking the mean 2. Use a 17-dimensional vector comprising of MSE (k), [S(k) ]2, R(states) (k) j, and S Ci for each i. Note that there is only one value for the third feature and 4 values for the fourth feature giving a total of ( = 17)-dimensional vector. In this case the fusion is achieved by dimensionality reduction using principal component analysis (PCA) retaining those combined features that capture 98% of the total variation. PCA is performed prior to classification. In both cases, a neural network classifier is used with one hidden layer of 6 nodes 2. Classification performance is analyzed for the following five cases, Case 1 Train on Building 1 data, test on Building 2 data Case 2 Train on Building 1 data, test on Building 1 data 3. Case 3 Train on Building 2 data, test on Building 2 data 3. Case 4 Train on Building 2 data, test on Building 1 data. Case 5 Mixed, i.e., combine data from both buildings by random permutation and use half the dataset to train and remaining half to test Tables 1 and 2 contain the results on test data; cases 2, 3 and 5 are the average of 1 iterations. We conclude that, in general, the classification performance is 2 We experimented with 3 to 9 nodes in the hidden layer. Best training performance was observed with 6 nodes. 3 Training is done using a randomly selected set of 9 trials out of the 12.

9 Table 1. Classification results: 4-dimensional feature vector. True/Error classes Case 1 Case 2 Case 3 Case 4 Case 5 C 1 true (T P ) C 2 true (T N) Type I error (F A) Type II error (M) Performance (P ) 87.5% 88.33% 95.% 77.5% 9.41% Table 2. Classification results: 17-dimensional feature vector followed by PCA. True/Error classes Case 1 Case 2 Case 3 Case 4 Case 5 C 1 true (T P ) C 2 true (T N) Type I error (F A) 3 6 Type II error (M) Performance (P ) 87.91% 88.33% 96.67% 78.75% 92.5% T P : True positives, T N: True negatives, F A: False alarms, M: Misses, P = (T P + T N)/(T P + T N + F A + M) 1% very good. Best performance is observed for Case 3 (Building 2). This indicates that selected features are good. If, somehow, we can improve the method of sensor data collection, then the classification performance will further improve. The good classification performance of the classifiers trained on the data from one building and tested on the same building supports the above conclusion. To assess the contribution of each feature alone we experimented with Case 5. It was observed that the third feature (R(states) (k) j ) contributes significantly to improving the classification performance (about 7%), whereas features MSE (k) and S Ci make similar amount of contributions. The similarity in performance between these two features is not surprising because they are both derived from correlation sequences. Since the rationale for using S Ci depends on the linear configuration of sensors, one may consider dropping this feature for a simpler system design under more general topology of sensor deployment. We have noticed that in spite of dropping this feature we get a classification performance of 89.13% for Case 5. 6 Concluding Remarks We have addressed the task of distinguishing between the presence of one vs. two persons in a visually unobservable indoor region of interest, using data obtained from seismic sensors. After exploring multiple alternatives, we identified four features as being capable of assisting classification with a high degree of reliability. Classification was achieved using a neural network classifier.

10 Although the current suite of four features has yielded very good performance, further improvements may be possible by exploring new feature extraction approaches. Performance gains may be obtained by using non-linear dimensionality reduction schemes [12] as opposed to principal component analysis, which is linear. Improved classification as well as finer classification is also expected using other modalities; such as the acoustic sensors. Since the data are vibrational in nature, the signal processing and feature extraction algorithms developed in this paper can also be applied to data collected using accelerometers and acoustic sensors. References 1. Dibazar, A.A., Park, H.O., Berger, T.W.: The application of dynamic synapse neural networks on footstep and vehicle recognition. In: Proc. International Joint Conference on Neural Networks IJCNN 27. (12 17 Aug. 27) Li, D., Wong, K.D., Hu, Y.H., Sayeed, A.M.: Detection, classification, and tracking of targets. IEEE Signal Processing Magazine 19(2) (22) Tian, Y., Qi, H., Wang, X.: Target detection and classification using seismic signal processing in unattended ground sensor systems. In: Proc. IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2). Volume 4. (22) IV Bland, R.E.: Acoustic and seismic signal processing for footstep detection. Master s thesis, Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science (26) 5. Succi, G., Clapp, D., Gampert, R., Prado, G.: Footstep detection and tracking. In: Proc. SPIE - Int. Soc. Opt. Eng. (USA). Volume 4393., USA (21) Subramanian, A., Iyengar, S.G., Mehrotra, K.G., Mohan, C.K., Varshney, P.K., Damarla, T.: A data-driven personnel detection scheme for indoor surveillance using seismic sensors. In Carapezza, E.M., ed.: Unattended Ground, Sea, and Air Sensor Technologies and Applications XI. Volume 7333., SPIE (29) Sabatier, J.M., Ekimov, A.E.: A review of human signatures in urban environments using seismic and acoustic methods. In: Proc. IEEE Conference on Technologies for Homeland Security. (12 13 May 28) Iyengar, S.G., Varshney, P.K., Damarla, T.: On the detection of footsteps based on acoustic and seismic sensing. In: Conference Record of the Forty-First Asilomar Conference on Signals, Systems and Computers ACSSC 27. (4 7 Nov. 27) Houston, K.M., McGaffigan, D.P.: Spectrum analysis techniques for personnel detection using seismic sensors. In Carapezza, E.M., ed.: Unattended Ground Sensor Technologies and Applications V. Volume 59., SPIE (23) Huang, N.E., Shen, Z., Long, S.R., Wu, M.C., Shih, H.H., Zheng, Q., Yen, N.C., Tung, C.C., Liu, H.H.: The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 454(1971) (1998) Vlachos, M., Yu, P.S., Castelli, V., Meek, C.: Structural periodic measures for time-series data. Data Mining and Knowledge Discovery 12(1) (Jan 26) Lee, J.A., Verleysen, M.: Nonlinear Dimensionality Reduction. Springer (27)

Empirical Mode Decomposition Based Denoising by Customized Thresholding

Empirical Mode Decomposition Based Denoising by Customized Thresholding Vol:11, No:5, 17 Empirical Mode Decomposition Based Denoising by Customized Thresholding Wahiba Mohguen, Raïs El hadi Bekka International Science Index, Electronics and Communication Engineering Vol:11,

More information

CHAPTER 3. Preprocessing and Feature Extraction. Techniques

CHAPTER 3. Preprocessing and Feature Extraction. Techniques CHAPTER 3 Preprocessing and Feature Extraction Techniques CHAPTER 3 Preprocessing and Feature Extraction Techniques 3.1 Need for Preprocessing and Feature Extraction schemes for Pattern Recognition and

More information

Visual object classification by sparse convolutional neural networks

Visual object classification by sparse convolutional neural networks Visual object classification by sparse convolutional neural networks Alexander Gepperth 1 1- Ruhr-Universität Bochum - Institute for Neural Dynamics Universitätsstraße 150, 44801 Bochum - Germany Abstract.

More information

Performance Degradation Assessment and Fault Diagnosis of Bearing Based on EMD and PCA-SOM

Performance Degradation Assessment and Fault Diagnosis of Bearing Based on EMD and PCA-SOM Performance Degradation Assessment and Fault Diagnosis of Bearing Based on EMD and PCA-SOM Lu Chen and Yuan Hang PERFORMANCE DEGRADATION ASSESSMENT AND FAULT DIAGNOSIS OF BEARING BASED ON EMD AND PCA-SOM.

More information

Multi-sensory Features for Personnel Detection at Border Crossing

Multi-sensory Features for Personnel Detection at Border Crossing Multi-sensory Features for Personnel Detection at Border Crossing Po-Sen Huang 1, Thyagaraju Damarla 2, Mark Hasegawa-Johnson 1 1 Beckman Institute, ECE Department, University of Illinois at Urbana-Champaign,

More information

An Improved Images Watermarking Scheme Using FABEMD Decomposition and DCT

An Improved Images Watermarking Scheme Using FABEMD Decomposition and DCT An Improved Images Watermarking Scheme Using FABEMD Decomposition and DCT Noura Aherrahrou and Hamid Tairi University Sidi Mohamed Ben Abdellah, Faculty of Sciences, Dhar El mahraz, LIIAN, Department of

More information

SUMMARY INTRODUCTION METHOD. Review of VMD theory

SUMMARY INTRODUCTION METHOD. Review of VMD theory Bin Lyu*, The University of Olahoma; Fangyu Li, The University of Georgia; Jie Qi, Tao Zhao, and Kurt J. Marfurt, The University of Olahoma SUMMARY The coherence attribute is a powerful tool to delineate

More information

Learning the Three Factors of a Non-overlapping Multi-camera Network Topology

Learning the Three Factors of a Non-overlapping Multi-camera Network Topology Learning the Three Factors of a Non-overlapping Multi-camera Network Topology Xiaotang Chen, Kaiqi Huang, and Tieniu Tan National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy

More information

Separation of Surface Roughness Profile from Raw Contour based on Empirical Mode Decomposition Shoubin LIU 1, a*, Hui ZHANG 2, b

Separation of Surface Roughness Profile from Raw Contour based on Empirical Mode Decomposition Shoubin LIU 1, a*, Hui ZHANG 2, b International Conference on Advances in Mechanical Engineering and Industrial Informatics (AMEII 2015) Separation of Surface Roughness Profile from Raw Contour based on Empirical Mode Decomposition Shoubin

More information

A Self-Organizing Binary System*

A Self-Organizing Binary System* 212 1959 PROCEEDINGS OF THE EASTERN JOINT COMPUTER CONFERENCE A Self-Organizing Binary System* RICHARD L. MATTSONt INTRODUCTION ANY STIMULUS to a system such as described in this paper can be coded into

More information

Spatial Outlier Detection

Spatial Outlier Detection Spatial Outlier Detection Chang-Tien Lu Department of Computer Science Northern Virginia Center Virginia Tech Joint work with Dechang Chen, Yufeng Kou, Jiang Zhao 1 Spatial Outlier A spatial data point

More information

SPARSE COMPONENT ANALYSIS FOR BLIND SOURCE SEPARATION WITH LESS SENSORS THAN SOURCES. Yuanqing Li, Andrzej Cichocki and Shun-ichi Amari

SPARSE COMPONENT ANALYSIS FOR BLIND SOURCE SEPARATION WITH LESS SENSORS THAN SOURCES. Yuanqing Li, Andrzej Cichocki and Shun-ichi Amari SPARSE COMPONENT ANALYSIS FOR BLIND SOURCE SEPARATION WITH LESS SENSORS THAN SOURCES Yuanqing Li, Andrzej Cichocki and Shun-ichi Amari Laboratory for Advanced Brain Signal Processing Laboratory for Mathematical

More information

Feature Selection Using Principal Feature Analysis

Feature Selection Using Principal Feature Analysis Feature Selection Using Principal Feature Analysis Ira Cohen Qi Tian Xiang Sean Zhou Thomas S. Huang Beckman Institute for Advanced Science and Technology University of Illinois at Urbana-Champaign Urbana,

More information

Data Mining Chapter 3: Visualizing and Exploring Data Fall 2011 Ming Li Department of Computer Science and Technology Nanjing University

Data Mining Chapter 3: Visualizing and Exploring Data Fall 2011 Ming Li Department of Computer Science and Technology Nanjing University Data Mining Chapter 3: Visualizing and Exploring Data Fall 2011 Ming Li Department of Computer Science and Technology Nanjing University Exploratory data analysis tasks Examine the data, in search of structures

More information

USING LINEAR PREDICTION TO MITIGATE END EFFECTS IN EMPIRICAL MODE DECOMPOSITION. Steven Sandoval, Matthew Bredin, and Phillip L.

USING LINEAR PREDICTION TO MITIGATE END EFFECTS IN EMPIRICAL MODE DECOMPOSITION. Steven Sandoval, Matthew Bredin, and Phillip L. USING LINEAR PREDICTION TO MITIGATE END EFFECTS IN EMPIRICAL MODE DECOMPOSITION Steven Sandoval, Matthew Bredin, and Phillip L. De Leon New Mexico State University Klipsch School of Electrical and Computer

More information

Rank Measures for Ordering

Rank Measures for Ordering Rank Measures for Ordering Jin Huang and Charles X. Ling Department of Computer Science The University of Western Ontario London, Ontario, Canada N6A 5B7 email: fjhuang33, clingg@csd.uwo.ca Abstract. Many

More information

Empirical Mode Decomposition: Improvement and Application

Empirical Mode Decomposition: Improvement and Application Empirical Mode Decomposition: Improvement and Application Peel, M.C. 1, G.G.S. Pegram 2 and T.A. McMahon 1 1 Department of Civil and Environmental Engineering, The University of Melbourne, Victoria 2 Civil

More information

MRF-based Algorithms for Segmentation of SAR Images

MRF-based Algorithms for Segmentation of SAR Images This paper originally appeared in the Proceedings of the 998 International Conference on Image Processing, v. 3, pp. 770-774, IEEE, Chicago, (998) MRF-based Algorithms for Segmentation of SAR Images Robert

More information

COSC160: Detection and Classification. Jeremy Bolton, PhD Assistant Teaching Professor

COSC160: Detection and Classification. Jeremy Bolton, PhD Assistant Teaching Professor COSC160: Detection and Classification Jeremy Bolton, PhD Assistant Teaching Professor Outline I. Problem I. Strategies II. Features for training III. Using spatial information? IV. Reducing dimensionality

More information

FMA901F: Machine Learning Lecture 3: Linear Models for Regression. Cristian Sminchisescu

FMA901F: Machine Learning Lecture 3: Linear Models for Regression. Cristian Sminchisescu FMA901F: Machine Learning Lecture 3: Linear Models for Regression Cristian Sminchisescu Machine Learning: Frequentist vs. Bayesian In the frequentist setting, we seek a fixed parameter (vector), with value(s)

More information

Design and Performance Improvements for Fault Detection in Tightly-Coupled Multi-Robot Team Tasks

Design and Performance Improvements for Fault Detection in Tightly-Coupled Multi-Robot Team Tasks Design and Performance Improvements for Fault Detection in Tightly-Coupled Multi-Robot Team Tasks Xingyan Li and Lynne E. Parker Distributed Intelligence Laboratory, Department of Electrical Engineering

More information

QUANTIZER DESIGN FOR EXPLOITING COMMON INFORMATION IN LAYERED CODING. Mehdi Salehifar, Tejaswi Nanjundaswamy, and Kenneth Rose

QUANTIZER DESIGN FOR EXPLOITING COMMON INFORMATION IN LAYERED CODING. Mehdi Salehifar, Tejaswi Nanjundaswamy, and Kenneth Rose QUANTIZER DESIGN FOR EXPLOITING COMMON INFORMATION IN LAYERED CODING Mehdi Salehifar, Tejaswi Nanjundaswamy, and Kenneth Rose Department of Electrical and Computer Engineering University of California,

More information

Image Quality Assessment Techniques: An Overview

Image Quality Assessment Techniques: An Overview Image Quality Assessment Techniques: An Overview Shruti Sonawane A. M. Deshpande Department of E&TC Department of E&TC TSSM s BSCOER, Pune, TSSM s BSCOER, Pune, Pune University, Maharashtra, India Pune

More information

ADVANCED IMAGE PROCESSING METHODS FOR ULTRASONIC NDE RESEARCH C. H. Chen, University of Massachusetts Dartmouth, N.

ADVANCED IMAGE PROCESSING METHODS FOR ULTRASONIC NDE RESEARCH C. H. Chen, University of Massachusetts Dartmouth, N. ADVANCED IMAGE PROCESSING METHODS FOR ULTRASONIC NDE RESEARCH C. H. Chen, University of Massachusetts Dartmouth, N. Dartmouth, MA USA Abstract: The significant progress in ultrasonic NDE systems has now

More information

EFFICIENT ADAPTIVE PREPROCESSING WITH DIMENSIONALITY REDUCTION FOR STREAMING DATA

EFFICIENT ADAPTIVE PREPROCESSING WITH DIMENSIONALITY REDUCTION FOR STREAMING DATA INTERNATIONAL JOURNAL OF RESEARCH IN COMPUTER APPLICATIONS AND ROBOTICS ISSN 2320-7345 EFFICIENT ADAPTIVE PREPROCESSING WITH DIMENSIONALITY REDUCTION FOR STREAMING DATA Saranya Vani.M 1, Dr. S. Uma 2,

More information

Detection, Classification, & Identification of Objects in Cluttered Images

Detection, Classification, & Identification of Objects in Cluttered Images Detection, Classification, & Identification of Objects in Cluttered Images Steve Elgar Washington State University Electrical Engineering 2752 Pullman, Washington 99164-2752 elgar@eecs.wsu.edu Voice: (509)

More information

The Study on Paper Board Thickness Measurement by Using Data Fusion

The Study on Paper Board Thickness Measurement by Using Data Fusion The Study on Paper Board Thickness Measurement by Using Data Fusion Lianhua Hu 1, Xin ping Li 1,2, Wei Tang 1, Qinghong Liu 3 1 School of Light Industry and Energy Shaanxi University of Science & Technology

More information

Adaptive Boundary Effect Processing For Empirical Mode Decomposition Using Template Matching

Adaptive Boundary Effect Processing For Empirical Mode Decomposition Using Template Matching Appl. Math. Inf. Sci. 7, No. 1L, 61-66 (2013) 61 Applied Mathematics & Information Sciences An International Journal Adaptive Boundary Effect Processing For Empirical Mode Decomposition Using Template

More information

Prediction of traffic flow based on the EMD and wavelet neural network Teng Feng 1,a,Xiaohong Wang 1,b,Yunlai He 1,c

Prediction of traffic flow based on the EMD and wavelet neural network Teng Feng 1,a,Xiaohong Wang 1,b,Yunlai He 1,c 2nd International Conference on Electrical, Computer Engineering and Electronics (ICECEE 215) Prediction of traffic flow based on the EMD and wavelet neural network Teng Feng 1,a,Xiaohong Wang 1,b,Yunlai

More information

Automatic Fatigue Detection System

Automatic Fatigue Detection System Automatic Fatigue Detection System T. Tinoco De Rubira, Stanford University December 11, 2009 1 Introduction Fatigue is the cause of a large number of car accidents in the United States. Studies done by

More information

An Optimized Embedded Target Detection System Using Acoustic and Seismic Sensors

An Optimized Embedded Target Detection System Using Acoustic and Seismic Sensors An Optimized Embedded Target Detection System Using Acoustic and Seismic Sensors Kyunghun Lee, Benjamin S. Riggan, and Shuvra S. Bhattacharyya Department of Electrical and Computer Engineering, University

More information

MULTI-VIEW TARGET CLASSIFICATION IN SYNTHETIC APERTURE SONAR IMAGERY

MULTI-VIEW TARGET CLASSIFICATION IN SYNTHETIC APERTURE SONAR IMAGERY MULTI-VIEW TARGET CLASSIFICATION IN SYNTHETIC APERTURE SONAR IMAGERY David Williams a, Johannes Groen b ab NATO Undersea Research Centre, Viale San Bartolomeo 400, 19126 La Spezia, Italy Contact Author:

More information

Unsupervised Learning : Clustering

Unsupervised Learning : Clustering Unsupervised Learning : Clustering Things to be Addressed Traditional Learning Models. Cluster Analysis K-means Clustering Algorithm Drawbacks of traditional clustering algorithms. Clustering as a complex

More information

On Finding Power Method in Spreading Activation Search

On Finding Power Method in Spreading Activation Search On Finding Power Method in Spreading Activation Search Ján Suchal Slovak University of Technology Faculty of Informatics and Information Technologies Institute of Informatics and Software Engineering Ilkovičova

More information

Human Gait Recognition Using Bezier Curves

Human Gait Recognition Using Bezier Curves Human Gait Recognition Using Bezier Curves Pratibha Mishra Samrat Ashok Technology Institute Vidisha, (M.P.) India Shweta Ezra Dhar Polytechnic College Dhar, (M.P.) India Abstract-- Gait recognition refers

More information

A Fast Personal Palm print Authentication based on 3D-Multi Wavelet Transformation

A Fast Personal Palm print Authentication based on 3D-Multi Wavelet Transformation A Fast Personal Palm print Authentication based on 3D-Multi Wavelet Transformation * A. H. M. Al-Helali, * W. A. Mahmmoud, and * H. A. Ali * Al- Isra Private University Email: adnan_hadi@yahoo.com Abstract:

More information

Research on the New Image De-Noising Methodology Based on Neural Network and HMM-Hidden Markov Models

Research on the New Image De-Noising Methodology Based on Neural Network and HMM-Hidden Markov Models Research on the New Image De-Noising Methodology Based on Neural Network and HMM-Hidden Markov Models Wenzhun Huang 1, a and Xinxin Xie 1, b 1 School of Information Engineering, Xijing University, Xi an

More information

MULTIVARIATE TEXTURE DISCRIMINATION USING A PRINCIPAL GEODESIC CLASSIFIER

MULTIVARIATE TEXTURE DISCRIMINATION USING A PRINCIPAL GEODESIC CLASSIFIER MULTIVARIATE TEXTURE DISCRIMINATION USING A PRINCIPAL GEODESIC CLASSIFIER A.Shabbir 1, 2 and G.Verdoolaege 1, 3 1 Department of Applied Physics, Ghent University, B-9000 Ghent, Belgium 2 Max Planck Institute

More information

Online Pose Classification and Walking Speed Estimation using Handheld Devices

Online Pose Classification and Walking Speed Estimation using Handheld Devices Online Pose Classification and Walking Speed Estimation using Handheld Devices Jun-geun Park MIT CSAIL Joint work with: Ami Patel (MIT EECS), Jonathan Ledlie (Nokia Research), Dorothy Curtis (MIT CSAIL),

More information

Evaluation of Model-Based Condition Monitoring Systems in Industrial Application Cases

Evaluation of Model-Based Condition Monitoring Systems in Industrial Application Cases Evaluation of Model-Based Condition Monitoring Systems in Industrial Application Cases S. Windmann 1, J. Eickmeyer 1, F. Jungbluth 1, J. Badinger 2, and O. Niggemann 1,2 1 Fraunhofer Application Center

More information

A Neural Network for Real-Time Signal Processing

A Neural Network for Real-Time Signal Processing 248 MalkofT A Neural Network for Real-Time Signal Processing Donald B. Malkoff General Electric / Advanced Technology Laboratories Moorestown Corporate Center Building 145-2, Route 38 Moorestown, NJ 08057

More information

Exemplar Selection Methods to Distinguish Human from Animal Footsteps

Exemplar Selection Methods to Distinguish Human from Animal Footsteps Exemplar Selection Methods to Distinguish Human from Animal Footsteps Po-Sen Huang, Mark Hasegawa-Johnson, Thyagaraju Damarla Beckman Institute, ECE Department, University of Illinois at Urbana-Champaign,

More information

Chapter 3. Speech segmentation. 3.1 Preprocessing

Chapter 3. Speech segmentation. 3.1 Preprocessing , as done in this dissertation, refers to the process of determining the boundaries between phonemes in the speech signal. No higher-level lexical information is used to accomplish this. This chapter presents

More information

Abstract. Introduction

Abstract. Introduction A COMPARISON OF SHEAR WAVE VELOCITIES OBTAINED FROM THE CROSSHOLE SEISMIC, SPECTRAL ANALYSIS OF SURFACE WAVES AND MULTIPLE IMPACTS OF SURFACE WAVES METHODS Patrick K. Miller, Olson Engineering, Wheat Ridge,

More information

An ICA based Approach for Complex Color Scene Text Binarization

An ICA based Approach for Complex Color Scene Text Binarization An ICA based Approach for Complex Color Scene Text Binarization Siddharth Kherada IIIT-Hyderabad, India siddharth.kherada@research.iiit.ac.in Anoop M. Namboodiri IIIT-Hyderabad, India anoop@iiit.ac.in

More information

Using Genetic Algorithms to Improve Pattern Classification Performance

Using Genetic Algorithms to Improve Pattern Classification Performance Using Genetic Algorithms to Improve Pattern Classification Performance Eric I. Chang and Richard P. Lippmann Lincoln Laboratory, MIT Lexington, MA 021739108 Abstract Genetic algorithms were used to select

More information

Time Series Clustering Ensemble Algorithm Based on Locality Preserving Projection

Time Series Clustering Ensemble Algorithm Based on Locality Preserving Projection Based on Locality Preserving Projection 2 Information & Technology College, Hebei University of Economics & Business, 05006 Shijiazhuang, China E-mail: 92475577@qq.com Xiaoqing Weng Information & Technology

More information

Face Hallucination Based on Eigentransformation Learning

Face Hallucination Based on Eigentransformation Learning Advanced Science and Technology etters, pp.32-37 http://dx.doi.org/10.14257/astl.2016. Face allucination Based on Eigentransformation earning Guohua Zou School of software, East China University of Technology,

More information

The Pennsylvania State University. The Graduate School. College of Engineering ONLINE LIVESTREAM CAMERA CALIBRATION FROM CROWD SCENE VIDEOS

The Pennsylvania State University. The Graduate School. College of Engineering ONLINE LIVESTREAM CAMERA CALIBRATION FROM CROWD SCENE VIDEOS The Pennsylvania State University The Graduate School College of Engineering ONLINE LIVESTREAM CAMERA CALIBRATION FROM CROWD SCENE VIDEOS A Thesis in Computer Science and Engineering by Anindita Bandyopadhyay

More information

Sampling PCA, enhancing recovered missing values in large scale matrices. Luis Gabriel De Alba Rivera 80555S

Sampling PCA, enhancing recovered missing values in large scale matrices. Luis Gabriel De Alba Rivera 80555S Sampling PCA, enhancing recovered missing values in large scale matrices. Luis Gabriel De Alba Rivera 80555S May 2, 2009 Introduction Human preferences (the quality tags we put on things) are language

More information

MetroPro Surface Texture Parameters

MetroPro Surface Texture Parameters MetroPro Surface Texture Parameters Contents ROUGHNESS PARAMETERS...1 R a, R q, R y, R t, R p, R v, R tm, R z, H, R ku, R 3z, SR z, SR z X, SR z Y, ISO Flatness WAVINESS PARAMETERS...4 W a, W q, W y HYBRID

More information

Unsupervised Change Detection in Remote-Sensing Images using Modified Self-Organizing Feature Map Neural Network

Unsupervised Change Detection in Remote-Sensing Images using Modified Self-Organizing Feature Map Neural Network Unsupervised Change Detection in Remote-Sensing Images using Modified Self-Organizing Feature Map Neural Network Swarnajyoti Patra, Susmita Ghosh Department of Computer Science and Engineering Jadavpur

More information

AIIA shot boundary detection at TRECVID 2006

AIIA shot boundary detection at TRECVID 2006 AIIA shot boundary detection at TRECVID 6 Z. Černeková, N. Nikolaidis and I. Pitas Artificial Intelligence and Information Analysis Laboratory Department of Informatics Aristotle University of Thessaloniki

More information

Detecting Harmful Hand Behaviors with Machine Learning from Wearable Motion Sensor Data

Detecting Harmful Hand Behaviors with Machine Learning from Wearable Motion Sensor Data Detecting Harmful Hand Behaviors with Machine Learning from Wearable Motion Sensor Data Lingfeng Zhang and Philip K. Chan Florida Institute of Technology, Melbourne, FL 32901 lingfeng2013@my.fit.edu, pkc@cs.fit.edu

More information

Research on Evaluation Method of Video Stabilization

Research on Evaluation Method of Video Stabilization International Conference on Advanced Material Science and Environmental Engineering (AMSEE 216) Research on Evaluation Method of Video Stabilization Bin Chen, Jianjun Zhao and i Wang Weapon Science and

More information

9.9 Coherent Structure Detection in a Backward-Facing Step Flow

9.9 Coherent Structure Detection in a Backward-Facing Step Flow 9.9 Coherent Structure Detection in a Backward-Facing Step Flow Contributed by: C. Schram, P. Rambaud, M. L. Riethmuller 9.9.1 Introduction An algorithm has been developed to automatically detect and characterize

More information

Assignment 2. Classification and Regression using Linear Networks, Multilayer Perceptron Networks, and Radial Basis Functions

Assignment 2. Classification and Regression using Linear Networks, Multilayer Perceptron Networks, and Radial Basis Functions ENEE 739Q: STATISTICAL AND NEURAL PATTERN RECOGNITION Spring 2002 Assignment 2 Classification and Regression using Linear Networks, Multilayer Perceptron Networks, and Radial Basis Functions Aravind Sundaresan

More information

Introduction to Pattern Recognition Part II. Selim Aksoy Bilkent University Department of Computer Engineering

Introduction to Pattern Recognition Part II. Selim Aksoy Bilkent University Department of Computer Engineering Introduction to Pattern Recognition Part II Selim Aksoy Bilkent University Department of Computer Engineering saksoy@cs.bilkent.edu.tr RETINA Pattern Recognition Tutorial, Summer 2005 Overview Statistical

More information

Advanced Digital Signal Processing Adaptive Linear Prediction Filter (Using The RLS Algorithm)

Advanced Digital Signal Processing Adaptive Linear Prediction Filter (Using The RLS Algorithm) Advanced Digital Signal Processing Adaptive Linear Prediction Filter (Using The RLS Algorithm) Erick L. Oberstar 2001 Adaptive Linear Prediction Filter Using the RLS Algorithm A complete analysis/discussion

More information

Accelerometer Gesture Recognition

Accelerometer Gesture Recognition Accelerometer Gesture Recognition Michael Xie xie@cs.stanford.edu David Pan napdivad@stanford.edu December 12, 2014 Abstract Our goal is to make gesture-based input for smartphones and smartwatches accurate

More information

Comparison of Sequence Matching Techniques for Video Copy Detection

Comparison of Sequence Matching Techniques for Video Copy Detection Comparison of Sequence Matching Techniques for Video Copy Detection Arun Hampapur a, Ki-Ho Hyun b and Ruud Bolle a a IBM T.J Watson Research Center, 3 Saw Mill River Road, Hawthorne, NY 1532, USA b School

More information

Detecting Salient Contours Using Orientation Energy Distribution. Part I: Thresholding Based on. Response Distribution

Detecting Salient Contours Using Orientation Energy Distribution. Part I: Thresholding Based on. Response Distribution Detecting Salient Contours Using Orientation Energy Distribution The Problem: How Does the Visual System Detect Salient Contours? CPSC 636 Slide12, Spring 212 Yoonsuck Choe Co-work with S. Sarma and H.-C.

More information

Effects of Surface Geology on Seismic Motion

Effects of Surface Geology on Seismic Motion MAXIMUM LIKELIHOOD PARAMETER ESTIMATION FOR SURFACE WAVES: APPLICATION TO AMBIENT VIBRATIONS Stefano Maranò and Donat Fäh Christoph Reller and Hans-Andrea Loeliger ETH Zurich ETH Zurich Swiss Seismological

More information

Image denoising in the wavelet domain using Improved Neigh-shrink

Image denoising in the wavelet domain using Improved Neigh-shrink Image denoising in the wavelet domain using Improved Neigh-shrink Rahim Kamran 1, Mehdi Nasri, Hossein Nezamabadi-pour 3, Saeid Saryazdi 4 1 Rahimkamran008@gmail.com nasri_me@yahoo.com 3 nezam@uk.ac.ir

More information

Interpolation artifacts and bidimensional ensemble empirical mode decomposition

Interpolation artifacts and bidimensional ensemble empirical mode decomposition Interpolation artifacts and bidimensional ensemble empirical mode decomposition Jiajun Han* University of Alberta, Edmonton, Alberta, Canada, hjiajun@ualberta.ca Mirko van der Baan University of Alberta,

More information

Data fusion and multi-cue data matching using diffusion maps

Data fusion and multi-cue data matching using diffusion maps Data fusion and multi-cue data matching using diffusion maps Stéphane Lafon Collaborators: Raphy Coifman, Andreas Glaser, Yosi Keller, Steven Zucker (Yale University) Part of this work was supported by

More information

Estimating Noise and Dimensionality in BCI Data Sets: Towards Illiteracy Comprehension

Estimating Noise and Dimensionality in BCI Data Sets: Towards Illiteracy Comprehension Estimating Noise and Dimensionality in BCI Data Sets: Towards Illiteracy Comprehension Claudia Sannelli, Mikio Braun, Michael Tangermann, Klaus-Robert Müller, Machine Learning Laboratory, Dept. Computer

More information

Payload Length and Rate Adaptation for Throughput Optimization in Wireless LANs

Payload Length and Rate Adaptation for Throughput Optimization in Wireless LANs Payload Length and Rate Adaptation for Throughput Optimization in Wireless LANs Sayantan Choudhury and Jerry D. Gibson Department of Electrical and Computer Engineering University of Califonia, Santa Barbara

More information

pyeemd Documentation Release Perttu Luukko

pyeemd Documentation Release Perttu Luukko pyeemd Documentation Release 1.3.1 Perttu Luukko August 10, 2016 Contents 1 Contents: 3 1.1 Installing pyeemd............................................ 3 1.2 Tutorial..................................................

More information

4.12 Generalization. In back-propagation learning, as many training examples as possible are typically used.

4.12 Generalization. In back-propagation learning, as many training examples as possible are typically used. 1 4.12 Generalization In back-propagation learning, as many training examples as possible are typically used. It is hoped that the network so designed generalizes well. A network generalizes well when

More information

Data mining. Classification k-nn Classifier. Piotr Paszek. (Piotr Paszek) Data mining k-nn 1 / 20

Data mining. Classification k-nn Classifier. Piotr Paszek. (Piotr Paszek) Data mining k-nn 1 / 20 Data mining Piotr Paszek Classification k-nn Classifier (Piotr Paszek) Data mining k-nn 1 / 20 Plan of the lecture 1 Lazy Learner 2 k-nearest Neighbor Classifier 1 Distance (metric) 2 How to Determine

More information

Nonlinear Noise Reduction

Nonlinear Noise Reduction Chapter 3 Nonlinear Noise Reduction Many natural and engineered systems generate nonlinear deterministic time series that are contaminated by random measurement or dynamic noise. Even clean time series

More information

CANCER PREDICTION USING PATTERN CLASSIFICATION OF MICROARRAY DATA. By: Sudhir Madhav Rao &Vinod Jayakumar Instructor: Dr.

CANCER PREDICTION USING PATTERN CLASSIFICATION OF MICROARRAY DATA. By: Sudhir Madhav Rao &Vinod Jayakumar Instructor: Dr. CANCER PREDICTION USING PATTERN CLASSIFICATION OF MICROARRAY DATA By: Sudhir Madhav Rao &Vinod Jayakumar Instructor: Dr. Michael Nechyba 1. Abstract The objective of this project is to apply well known

More information

Anomaly Detection on Data Streams with High Dimensional Data Environment

Anomaly Detection on Data Streams with High Dimensional Data Environment Anomaly Detection on Data Streams with High Dimensional Data Environment Mr. D. Gokul Prasath 1, Dr. R. Sivaraj, M.E, Ph.D., 2 Department of CSE, Velalar College of Engineering & Technology, Erode 1 Assistant

More information

Application of Pattern Recognition for Damage Classification in Composite Laminates

Application of Pattern Recognition for Damage Classification in Composite Laminates Application of Pattern Recognition for Damage Classification in Composite Laminates Seth S. Kessler, Ph.D. Pramila Agrawal, Ph.D. IWSHM 10 Conference Canal Park 2007 Cambridge, MA 02141 MDC Proprietary

More information

Face Recognition Based on LDA and Improved Pairwise-Constrained Multiple Metric Learning Method

Face Recognition Based on LDA and Improved Pairwise-Constrained Multiple Metric Learning Method Journal of Information Hiding and Multimedia Signal Processing c 2016 ISSN 2073-4212 Ubiquitous International Volume 7, Number 5, September 2016 Face Recognition ased on LDA and Improved Pairwise-Constrained

More information

Rolling element bearings fault diagnosis based on CEEMD and SVM

Rolling element bearings fault diagnosis based on CEEMD and SVM Rolling element bearings fault diagnosis based on CEEMD and SVM Tao-tao Zhou 1, Xian-ming Zhu 2, Yan Liu 3, Wei-cai Peng 4 National Key Laboratory on Ship Vibration and Noise, China Ship Development and

More information

Audio Watermarking using Colour Image Based on EMD and DCT

Audio Watermarking using Colour Image Based on EMD and DCT Audio Watermarking using Colour Image Based on EMD and Suhail Yoosuf 1, Ann Mary Alex 2 P. G. Scholar, Department of Electronics and Communication, Mar Baselios College of Engineering and Technology, Trivandrum,

More information

Collaborative Filtering using a Spreading Activation Approach

Collaborative Filtering using a Spreading Activation Approach Collaborative Filtering using a Spreading Activation Approach Josephine Griffith *, Colm O Riordan *, Humphrey Sorensen ** * Department of Information Technology, NUI, Galway ** Computer Science Department,

More information

Classification of Subject Motion for Improved Reconstruction of Dynamic Magnetic Resonance Imaging

Classification of Subject Motion for Improved Reconstruction of Dynamic Magnetic Resonance Imaging 1 CS 9 Final Project Classification of Subject Motion for Improved Reconstruction of Dynamic Magnetic Resonance Imaging Feiyu Chen Department of Electrical Engineering ABSTRACT Subject motion is a significant

More information

Expectation and Maximization Algorithm for Estimating Parameters of a Simple Partial Erasure Model

Expectation and Maximization Algorithm for Estimating Parameters of a Simple Partial Erasure Model 608 IEEE TRANSACTIONS ON MAGNETICS, VOL. 39, NO. 1, JANUARY 2003 Expectation and Maximization Algorithm for Estimating Parameters of a Simple Partial Erasure Model Tsai-Sheng Kao and Mu-Huo Cheng Abstract

More information

A Data Classification Algorithm of Internet of Things Based on Neural Network

A Data Classification Algorithm of Internet of Things Based on Neural Network A Data Classification Algorithm of Internet of Things Based on Neural Network https://doi.org/10.3991/ijoe.v13i09.7587 Zhenjun Li Hunan Radio and TV University, Hunan, China 278060389@qq.com Abstract To

More information

Hybrid Face Recognition and Classification System for Real Time Environment

Hybrid Face Recognition and Classification System for Real Time Environment Hybrid Face Recognition and Classification System for Real Time Environment Dr.Matheel E. Abdulmunem Department of Computer Science University of Technology, Baghdad, Iraq. Fatima B. Ibrahim Department

More information

Image retrieval based on bag of images

Image retrieval based on bag of images University of Wollongong Research Online Faculty of Informatics - Papers (Archive) Faculty of Engineering and Information Sciences 2009 Image retrieval based on bag of images Jun Zhang University of Wollongong

More information

Statistics of Natural Image Categories

Statistics of Natural Image Categories Statistics of Natural Image Categories Authors: Antonio Torralba and Aude Oliva Presented by: Sebastian Scherer Experiment Please estimate the average depth from the camera viewpoint to all locations(pixels)

More information

Supplementary Figure 1. Decoding results broken down for different ROIs

Supplementary Figure 1. Decoding results broken down for different ROIs Supplementary Figure 1 Decoding results broken down for different ROIs Decoding results for areas V1, V2, V3, and V1 V3 combined. (a) Decoded and presented orientations are strongly correlated in areas

More information

Semi-Supervised PCA-based Face Recognition Using Self-Training

Semi-Supervised PCA-based Face Recognition Using Self-Training Semi-Supervised PCA-based Face Recognition Using Self-Training Fabio Roli and Gian Luca Marcialis Dept. of Electrical and Electronic Engineering, University of Cagliari Piazza d Armi, 09123 Cagliari, Italy

More information

Network Traffic Measurements and Analysis

Network Traffic Measurements and Analysis DEIB - Politecnico di Milano Fall, 2017 Introduction Often, we have only a set of features x = x 1, x 2,, x n, but no associated response y. Therefore we are not interested in prediction nor classification,

More information

Optimizing the Deblocking Algorithm for. H.264 Decoder Implementation

Optimizing the Deblocking Algorithm for. H.264 Decoder Implementation Optimizing the Deblocking Algorithm for H.264 Decoder Implementation Ken Kin-Hung Lam Abstract In the emerging H.264 video coding standard, a deblocking/loop filter is required for improving the visual

More information

Assignment 2. Unsupervised & Probabilistic Learning. Maneesh Sahani Due: Monday Nov 5, 2018

Assignment 2. Unsupervised & Probabilistic Learning. Maneesh Sahani Due: Monday Nov 5, 2018 Assignment 2 Unsupervised & Probabilistic Learning Maneesh Sahani Due: Monday Nov 5, 2018 Note: Assignments are due at 11:00 AM (the start of lecture) on the date above. he usual College late assignments

More information

DEVELOPMENT OF SUBSTRUCTURED SHAKING TABLE TEST METHOD

DEVELOPMENT OF SUBSTRUCTURED SHAKING TABLE TEST METHOD DEVELOPMENT OF SUBSTRUCTURED SHAKING TABLE TEST METHOD Akira IGARASHI 1, Hirokazu IEMURA 2 And Takanori SUWA 3 SUMMARY Since various kinds of issues arise in the practical application of structural response

More information

Data Mining. CS57300 Purdue University. Bruno Ribeiro. February 1st, 2018

Data Mining. CS57300 Purdue University. Bruno Ribeiro. February 1st, 2018 Data Mining CS57300 Purdue University Bruno Ribeiro February 1st, 2018 1 Exploratory Data Analysis & Feature Construction How to explore a dataset Understanding the variables (values, ranges, and empirical

More information

Image Quality Assessment based on Improved Structural SIMilarity

Image Quality Assessment based on Improved Structural SIMilarity Image Quality Assessment based on Improved Structural SIMilarity Jinjian Wu 1, Fei Qi 2, and Guangming Shi 3 School of Electronic Engineering, Xidian University, Xi an, Shaanxi, 710071, P.R. China 1 jinjian.wu@mail.xidian.edu.cn

More information

Applying Supervised Learning

Applying Supervised Learning Applying Supervised Learning When to Consider Supervised Learning A supervised learning algorithm takes a known set of input data (the training set) and known responses to the data (output), and trains

More information

Feature-level Fusion for Effective Palmprint Authentication

Feature-level Fusion for Effective Palmprint Authentication Feature-level Fusion for Effective Palmprint Authentication Adams Wai-Kin Kong 1, 2 and David Zhang 1 1 Biometric Research Center, Department of Computing The Hong Kong Polytechnic University, Kowloon,

More information

An Abnormal Data Detection Method Based on the Temporal-spatial Correlation in Wireless Sensor Networks

An Abnormal Data Detection Method Based on the Temporal-spatial Correlation in Wireless Sensor Networks An Based on the Temporal-spatial Correlation in Wireless Sensor Networks 1 Department of Computer Science & Technology, Harbin Institute of Technology at Weihai,Weihai, 264209, China E-mail: Liuyang322@hit.edu.cn

More information

Short Survey on Static Hand Gesture Recognition

Short Survey on Static Hand Gesture Recognition Short Survey on Static Hand Gesture Recognition Huu-Hung Huynh University of Science and Technology The University of Danang, Vietnam Duc-Hoang Vo University of Science and Technology The University of

More information

Performance Evaluation Metrics and Statistics for Positional Tracker Evaluation

Performance Evaluation Metrics and Statistics for Positional Tracker Evaluation Performance Evaluation Metrics and Statistics for Positional Tracker Evaluation Chris J. Needham and Roger D. Boyle School of Computing, The University of Leeds, Leeds, LS2 9JT, UK {chrisn,roger}@comp.leeds.ac.uk

More information

Function approximation using RBF network. 10 basis functions and 25 data points.

Function approximation using RBF network. 10 basis functions and 25 data points. 1 Function approximation using RBF network F (x j ) = m 1 w i ϕ( x j t i ) i=1 j = 1... N, m 1 = 10, N = 25 10 basis functions and 25 data points. Basis function centers are plotted with circles and data

More information

Video shot segmentation using late fusion technique

Video shot segmentation using late fusion technique Video shot segmentation using late fusion technique by C. Krishna Mohan, N. Dhananjaya, B.Yegnanarayana in Proc. Seventh International Conference on Machine Learning and Applications, 2008, San Diego,

More information