ASYMPTOTIC CLOSED-LOOP DESIGN OF ERROR RESILIENT PREDICTIVE COMPRESSION SYSTEMS. Sina Zamani, Tejaswi Nanjundaswamy, Kenneth Rose

Size: px
Start display at page:

Download "ASYMPTOTIC CLOSED-LOOP DESIGN OF ERROR RESILIENT PREDICTIVE COMPRESSION SYSTEMS. Sina Zamani, Tejaswi Nanjundaswamy, Kenneth Rose"

Transcription

1 ASYMPTOTIC OSED-LOOP DESIGN OF ERROR RESILIENT PREDICTIVE COMPRESSION SYSTEMS Sia Zamai, Tejaswi Najudaswamy, Keeth Rose Departmet of Electrical ad Computer Egieerig, Uiversity of Califoria Sata Barbara, CA {siazm, tejaswi, ABSTRACT Predictio is used i virtually all compressio systems. Whe such a compressed sigal is trasmitted over ureliable etworks, packet losses ca lead to sigificat error propagatio through the predictio loop. Despite this, the covetioal desig techique completely igores the effect of packet losses, ad estimates the predictio parameters to miimize the mea squared predictio error, ad optimizes the quatizer to miimize the recostructio error at the ecoder. While some desig techiques have bee proposed to accurately estimate ad miimize the ed-to-ed distortio at the decoder that accouts for packet losses, they operate i a closed-loop, which itroduces a mismatch betwee statistics used for desig ad statistics used i operatio, causig a egative impact o covergece ad stability of the desig procedure. Istead, we propose i this paper a effective techique for predictive compressio system desig that accouts for the istability caused by error propagatio due to packet losses, ad ejoys stable statistics durig desig by employig ope-loop iteratios that o covergece mimic closedloop operatio. Simulatio results for a compressio system with a first order liear predictor demostrate the utility of the proposed approach, which offers sigificat performace improvemets over eistig desig techiques. Ide Terms Predictor desig, Asymptotic Closed-Loop desig, Error Resiliece, Liear Predictor. INTRODUCTION Liear predictio is widely used i speech codig, speech sythesis, speech recogitio, audio codig, ad video codig. I compressio systems, the predictio module plays a importat role i eploitig temporal ad spatial redudacies. However, whe such a compressed data is trasmitted over ureliable etworks, errors itroduced due to the ievitable packet losses, propagate through the predictio loop, causig substatial, ad sometimes catastrophic, deterioratio of the received sigal. Despite this, covetioal compressio system desig completely igores the effect of chael loss, ad chooses the predictio parameters to miimize the mea squared predictio error, ad optimizes the quatizer to miimize the recostructio error at the ecoder. This problem ca be alleviated by optimizig the system for the overall ed-to-ed distortio (EED) observed at the decoder, which accouts for the effect of packet loss. A optimal recursive techique to estimate EED at the ecoder was proposed i [], ad utilizig this distortio to optimally select the parameters for motio compesated predictio i video codig was proposed i []. However, desigig optimal predictors ad quatizers, while accoutig for EED is a challegig task, as we eed to This work was supported i part by a gift from Mozilla. work with a stable traiig set that accurately represets the true sigal statistics. The ope-loop (OL) ad closed-loop () approaches were proposed i [3] for predictive vector quatizatio ad have bee widely used sice the. The OL approach uses the origial data as the predictio referece durig desig, but sice the decoder does ot have access to the origial data, the parameters desiged are ot suitable for the statistics see at the decoder. The approach attempts to alleviates the problem of this mismatch by desigig the parameters usig recostructed data obtaied i a closed-loop system as the predictio referece. However, usig these parameters i a closed-loop system geerates ew predictio ad recostructio, which implies they differ from the data the parameters were desiged for. This mismatch i statistics betwee desig ad operatio grows over time as the data is fed through the predictio loop i the coder, leadig to istability i both estimatio of predictio parameters, ad desig of quatizers, especially at lower bit rates. Note that this error propagatio ecoutered durig the desig phase due to statistics mismatch, differs from the error propagatio due to packet losses. I this paper we propose to address the challegig problem of tacklig these two types of error propagatio, by desigig the system iteratively, wherei a estimated EED is miimized at each iteratio to accout for packet losses, ad the predictio referece from the previous iteratio is employed i a ope-loop way to esure statistics used for desig ad operatio are matched. Oce the parameters beig desiged coverge, the predictio referece i the curret iteratio will match the referece from the previous iteratio, thus mimickig closed-loop operatio. Hece, we call this the asymptotic closed-loop () approach, which is similar to the approach i [, 5], wherei system desig without accoutig for packet losses is proposed. We specifically describe a framework for rate versus EED optimizatio of a compressio system employig a first order liear predictor. We also propose a ew ecoder architecture, i which the predictio at ecoder is based o the epected decoder recostructios. Eperimetal results substatiate the utility of the proposed approach with sigificat performace improvemets over eistig desig techiques.. PROBLEM SETUP Fig. illustrates a predictive compressio system, wherei iput sigal samples,, 0 < N, are coded by the ecoder to geerate a bitstream, which is trasmitted through a chael to the decoder, where it is decoded to geerate the recostructed samples. The ecoder uses its previous recostructed samples, ˆ e, to geerate predicted samples, e, ad the predictio error, e = e,. This is quatized to geerate ê, ad set over the chael. Whe the decoder receives ê, it adds it to its predicted sample, d, to geerate its recostructed samples, ˆ d,. Note that the recostructed samples //$ IEEE 3 ICASSP 0

2 Ecoder Decoder ˆ d, e ê Q E Chael ˆ (i ) 0 P (i ) ˆ (i ) Σ P (i ) Σ ˆ (i ) e, d, P D Desig a ew Quatizer Q (i) P E ˆ e, Fig. : A predictive compressio system 0 P (i ) Σ Q (i ) e (i) P (i ) ê (i) Σ Q (i) e (i) P (i ) Σ Q (i) ˆ (i) 0 P (i ) Σ Q (i ) Desig a ew Quatizer Q (i) ˆ (i) Desig a ew Predictor P (i) Fig. : Closed-loop traiig approach ê (i) at the ecoder, ˆ e,, ad the decoder, ˆ d,, will differ whe the chael is ureliable ad packets carryig ê are lost. This ucertaity results i ˆ d, beig a radom variable to the ecoder. The problem at had is to desig optimal quatizers ad predictors (P E, Q E ad P D) to miimize the epected EED at the decoder to accout for packet losses. For the mea squared error distortio metric, epected EED at the decoder is, E{D} = = E{( ˆ d, ) } ˆ (i) E{ˆ d, } E{(ˆ d, ) }. () Clearly, to estimate this distortio, first ad secod momets of the decoder recostructios should be accurately estimated at the ecoder. 3. BACKGROUND 3.. Ed to Ed distortio estimatio ad predictio A recursive techique to optimally estimate the epected EED at the ecoder i the presece of packet losses via the first ad secod momets of the decoder recostructios was proposed i [] for video coders. The recursive algorithm optimally estimates the decoder recostructios first momet, E{ˆ j d, }, ad the secod momet, E{(ˆ j d, ) }, for every piel j i frame. These momets are the used to estimate EED at the ecoder usig () to optimally switch betwee iter-frame predictio ad itra-frame predictio, to cotrol the error propagatio through frames. I [], a ew predictio scheme is employed i cojuctio with optimal EED estimatio. Q (i) Q (i) Desig a ew Predictor P (i) Fig. 3: Asymptotic closed-loop traiig approach Covetioal motio compesated predictio employs the ecoder recostructios for predictio, i.e., j e, = ˆ jv e,, where v is the optimal motio vector that miimizes the predictio error. Istead i [], the predictio is based o epected decoder recostructios, i.e., j e, = E{ˆ jv d, }, where v is the optimal motio vector that miimizes the EED of (). This setup plays a importat role i limitig error propagatio durig decoder operatio by appropriately selectig motio vectors to predict from referece blocks that are less likely to be corrupted by error propagatio. 3.. Closed-Loop versus Asymptotic Closed-Loop Desig I closed-loop iterative desig [], the coder operates i closedloop at each iteratio to geerate predictio errors ad recostructed samples that are used to desig the updated quatizer ad the updated predictor, respectively. At iteratio i, give a quatizer, Q (i ), ad a predictor, P (i ), a traiig set of predictio errors, T (i) : { } N =, for iteratio i is geerated as, where, = P (i ) ( ), () = P (i ) ( ) Q(i ) ( P (i ) ( )). (3) These two equatios are calculated sequetially for all values. The give T (i), we desig a ew quatizer, Q (i). Usig Q (i), a ew set of recostructed samples, ˆ (i), is geerated as per (3) ad based o this, we desig a ew predictor, P (i). These steps are repeated util covergece. Fig. illustrates this closed-loop iterative desig. The major issue with this approach is that whe the updated parameters are employed i closed-loop at a iteratio, ew predictio errors are geerated, which differ from the errors the quatizer was desiged for, ad this implies differet recostructios are geerated, which differ from the referece recostructios the predictor was desiged for. This mismatch i statistics betwee desig ad operatio builds up over time as the data is fed through the predictio loop i the coder, leadig to istability i the iterative desig of both the predictor ad the quatizer, especially at lower bit rates. The desig techique proposed i [, 5], tackles this statistics mismatch issue by desigig the predictor ad the quatizer i a opeloop fashio, while ultimately optimizig the system for closed-loop operatio. Specifically, the predictio is based o recostructios of previous iteratio, i.e.,the predictio errors are geerated as, = P (i ) (ˆ (i ) ). () 3

3 Give the ew predictio errors, we desig a ew quatizer, Q (i). This Q (i) is ow employed to geerate the recostructed samples of et iteratio as, = P (i ) (ˆ (i ) ) Q(i) ( P (i ) (ˆ (i ) )), (5) agai usig the recostructios of previous iteratio for predictio. Give these ew recostructios, we desig a ew predictor, P (i). Note that the equatios () ad (5) are eecuted idepedetly for each sample of the sequece i a ope-loop way. The mai steps of this techique are depicted i Fig. 3. The ope-loop format esures the predictor ad quatizer employ eactly the same recostructed data ad predictio error used for their desig, elimiatig the statistical mismatch issue see i closed-loop desig. O covergece, the predictor ad the quatizer do ot chage, which implies,, i.e., predictig from previous iteratio recostructios is the same as predictig from the curret iteratio recostructios, which is effectively closed-loop operatio. = ˆ(i ). PROPOSED APPROACH We propose a framework to desig a first order predictor ad a quatizer to miimize the EED i (). We first develop the EED estimatio algorithm, the we propose a ecoder architecture i which predictios are based o the epected recostructios at the decoder ad fially we propose the desig approach that accouts for packet loss to improve codig efficiecy ad desig stability... Epected Decoder Distortio ad Recostructios We assume for simplicity of presetatio that each packet cotais oe sample (or alteratively that iterleavig is used). The packet (or sample) loss rate is deoted as p. The predictio model employed at the decoder is a simple first order liear predictor, d, = αˆ d,, () where α is the predictio coefficiet that eeds to be estimated. The quatized predictio error, ê, trasmitted over the chael, may or may ot be received by the decoder. If the curret packet is received (with probability p), the decoder uses it to geerate the recostructed sample as, ˆ d, = d, ê. (7) Whe the packet is lost, a simple cocealmet of settig residue to zero is employed, which gives the recostructed sample as, ˆ d, = d,. () Thus the first ad secod momet of the decoder recostructed samples, required to estimate EED give i (), are calculated recursively at the ecoder as, E{ˆ d, } = ( p)e{ê αˆ d, } pe{αˆ d, } = ( p)ê αe{ˆ d, } (9) E{(ˆ d, ) } = ( p)e{(ê αˆ d, ) } pe{(αˆ d, ) } = ( p)(ê αê E{ˆ d, }) α E{(ˆ d, ) } (0) e ê Σ Q Chael e, = αe{ˆ d, } Ecoder P ( p)ê E{ˆ d, } D d, = αˆ d, Fig. : Architecture of the proposed coder Decoder.. Predictio Based o the Epected Decoder Recostructios Packet losses cause the recostructios at the ecoder ad the decoder to differ. Thus to close the gap betwee predictio at the ecoder ad the decoder, we employ the epected decoder recostructios for predictio at the ecoder, i.e., P ˆ d, e, = αe{ˆ d, }. () The predictio error, e = e,, is the quatized to geerate, ê. The overall proposed architecture is show i Fig.. We desig the predictio coefficiet α to miimize the EED, by solvig for α i the equatio give by settig the partial derivative of EED with respect to α to 0. The EED i () is depedet o α through equatios (9) ad (0). The equatio to be solved is, E{D} α = E{ˆ d, } ( p)ê E{ˆ d, } αe{(ˆ d, ) } = 0, () which gives us the solutio as, α = E{ˆ d, }( ( p)ê ). (3) E{(ˆ d, ) } Note that although ê is depedet o α, we assume that the modificatios i α across our desig iteratios are small eough to ot chage the quatizatio itervals..3. Asymptotic Closed-Loop Desig We employ the approach for a stable system desig by elimiatig the statistical mismatch issue of closed-loop desig. This is achieved by operatig i a ope-loop way, wherei we employ previous iteratio s first ad secod momets of the decoder recostructios, to estimate curret iteratio s momets ad predictio. Give a set of decoder recostructios first momets, E{ˆ d } (i ), ad secod momets, E{(ˆ d ) } (i ), of iteratio i, the predictor ad quatizer are iteratively desiged i a ier loop. I a subiteratio s of the ier loop, give a set of quatized predictio errors, ê (i,s ), the optimal predictio coefficiet is estimated as, α (i,s) = E{ˆ d, } (i ) ( ( p)ê (i,s ) ) E{(ˆ d, ) } (i ). () D 33

4 -ER -ER -ER (a) (b) (c) Fig. 5: Average decoder distortio (i db) of, ad the proposed -ER desig approaches, for a first order predictive coder with etropy costraied scalar quatizer, at various bit rates for packet loss rates of (a)5% (b)0% (c)0%. The ew predictio errors are ow geerated i a ope-loop fashio as, e (i,s) = α (i,s) E{ˆ d, } (i ). (5) A optimal quatizer, Q (i,s), is ow desiged for this set of ew predictio errors, which is used to geerate a ew set of quatized predictio errors, ê (i,s) = Q (i,s) (e (i,s) ). These subiteratios are repeated util covergece to obtai curret subiteratios fial quatizer, Q (i), fial predictio coefficiet, α (i), ad fial set of quatized predictio errors,. The first ad secod momets of the decoder recostructios are ow updated i the outer loop i a opeloop way as, E{ˆ d, } (i) E{(ˆ d, ) } (i) = ( p) α (i) E{ˆ d, } (i ) () = ( p)(( ) α (i) E{ˆ d, } (i ) ) (α (i) ) E{(ˆ d, ) } (i ). (7) These momets are ow used i the et iteratios ier loop to update the predictor ad quatizer. Iteratios are repeated util covergece. Note that although the etire desig is i ope-loop, o covergece it emulates closed-loop operatio. This is achieved as o covergece the quatizer ad predictor do ot chage, i.e., Q (i) = Q (i ) ad α (i) = α (i ), which implies E{ˆ d, } (i) = E{ˆ d, } (i ), thus employig previous iteratio s momets is the same as estimatig curret iteratio s momets recursively ad employig them for predictio i a closed-loop way. 5. EXPERIMENTAL RESULTS To validate our proposed method, we evaluated it for a compressio system with first order liear predictio ad a etropy costraied scalar quatizer. The Geeralized Lloyd Algorithm (GLA) was used to desig the etropy costraied scalar quatizer. We used the speech files available i the EBU SQAM database [7] as our dataset, as liear predictio is commoly employed i speech codig. However, ote that the proposed approach is applicable to predictive compressio of ay sigal with temporal correlatios. The first half of the speech files were used as traiig set (resultig i more millio samples) ad the secod half as test data. The predictio coefficiet was iitialized to zero. We evaluated the followig three differet desig techiques:. The closed-loop desig procedure discussed i Sectio 3., which completely igores the packet losses (referred as ).. The algorithm discussed i [5], which also igores the packet losses ad ca be obtaied by settig p = 0 i the update formulas of Sectio.3 (referred as ). 3. Our proposed method (referred as -ER). The system performace is evaluated by plottig the decoder recostructio error s sigal to oise ratio () averaged over te differet loss patters versus average bitrate, as show i Fig. 5 for differet packet loss rates. Clearly, the proposed approach cosistetly outperforms both ad uder all testig scearios, with gais of up to 7 db over ad gais of up to.5 db over. The proposed approach provides higher performace improvemets as the packet loss rate icreases, sice accoutig for error propagatio due to packet losses becomes critical i these cases. Compared to, our method provides larger gais at higher bitrates, as relies more o the high quality previous recostructios for predictio, uaware of the fact that these recostructios at the decoder will be corrupted as a result of error propagatio due to packet losses. These results substatiate the sigificat utility of the proposed approach.. CONUSION I this paper we proposed a effective ad robust desig techique for a predictive compressio system. We accout for the presece of a ureliable chael by desigig the system to optimize the EED at the decoder. We the elimiate the statistical mismatch issue suffered by covetioal closed-loop approaches, by employig a stable iterative desig approach that operates i a ope-loop way ad o covergece mimics closed-loop operatio. By carefully desigig the system parameters, error propagatio at the decoder is effectively cotaied. Sigificat performace improvemets see i eperimetal evaluatio results demostrate the utility of the proposed approach. Future research directios iclude, etedig the proposed desig techique to higher order predictors, ad employig a powerful optimizatio techique for desig of the etropy costraied scalar quatizer to accout for packet losses. 7. REFERENCES [] R. Zhag, S. L. Reguatha, ad K. Rose, Video codig with optimal iter/itra-mode switchig for packet loss resiliece, 3

5 IEEE Joural o Selected Areas i Commuicatios, vol., o., pp. 9 97, 000. [] H. Yag ad K. Rose, Source-chael predictio i error resiliet video codig, i IEEE Iteratioal Coferece o Multimedia ad Epo (ICME), 003, vol., pp. II 33. [3] V. Cuperma ad A. Gersho, Vector predictive codig of speech at kbits/s, IEEE Trasactios o Commuicatios, vol. 33, o. 7, pp. 5 9, 95. [] H. Khalil, K. Rose, ad S. L. Reguatha, The asymptotic closed-loop approach to predictive vector quatizer desig with applicatio i video codig, IEEE Trasactios o Image Processig, vol. 0, o., pp. 5 3, 00. [5] H. Khalil ad K. Rose, Robust predictive vector quatizer desig, i Data Compressio Coferece (DCC), 00, pp. 33. [] P.-C Chag ad R. M. Gray, Gradiet algorithms for desigig predictive vector quatizers, IEEE Trasactios o Acoustics, Speech ad Sigal Processig, vol. 3, o., pp , 9. [7] G Waters, Soud quality assessmet materialrecordigs for subjective tests: Users hadbook for the ebu sqam compact disk, Europea Broadcastig Uio (EBU), Tech. Rep, 9. 35

Run-length & Entropy Coding. Redundancy Removal. Sampling. Quantization. Perform inverse operations at the receiver EEE

Run-length & Entropy Coding. Redundancy Removal. Sampling. Quantization. Perform inverse operations at the receiver EEE Geeral e Image Coder Structure Motio Video (s 1,s 2,t) or (s 1,s 2 ) Natural Image Samplig A form of data compressio; usually lossless, but ca be lossy Redudacy Removal Lossless compressio: predictive

More information

Packet Video 99. A Corruption Model for Motion Compensated Video Subject to Bit Errors. Gustavo de los Reyes Amy R. Reibman Shih-Fu Chang

Packet Video 99. A Corruption Model for Motion Compensated Video Subject to Bit Errors. Gustavo de los Reyes Amy R. Reibman Shih-Fu Chang A Corruptio Model for Motio Compesated Video Subject to Bit Errors Gustavo de los Reyes Amy R. Reibma Shih-Fu Chag AT&T Labs AT&T Labs-Research Columbia Uiversity gdelosreyes@att.com amy@research.att.com

More information

Digital Video Systems ECE 634

Digital Video Systems ECE 634 Digital Video Systems ECE 634 egieerig.purdue.edu/~reibma/ece634/idex.html Professor Amy Reibma MSEE 356 reibma@purdue.edu MulGple descripgo codig Layered codig vs. MDC Layered codig: Base layer is high

More information

Encoding-Assisted Temporal Direct Mode Decision for B Pictures in H.264/AVC

Encoding-Assisted Temporal Direct Mode Decision for B Pictures in H.264/AVC Ecodig-Assisted Temporal Direct Mode Decisio for B Pictures i H.64/AVC Ya-Neg Fag ad Yiyi Li Hui-Jae Hsieh Departmet of Commuicatio Egieerig Departmet of Iformatio Maagemet Natioal Cetral Uiversity, Taiwa

More information

Information-based Feature Selection

Information-based Feature Selection Iformatio-based Feature Selectio Farza Faria, Abbas Kazeroui, Afshi Babveyh Email: {faria,abbask,afshib}@staford.edu 1 Itroductio Feature selectio is a topic of great iterest i applicatios dealig with

More information

Vector Quantization: a Limiting Case of EM

Vector Quantization: a Limiting Case of EM . Itroductio & defiitios Assume that you are give a data set X = { x j }, j { 2,,, }, of d -dimesioal vectors. The vector quatizatio (VQ) problem requires that we fid a set of prototype vectors Z = { z

More information

Error Resilience Analysis of Multi-Hypothesis Motion Compensated Prediction for Video Coding

Error Resilience Analysis of Multi-Hypothesis Motion Compensated Prediction for Video Coding 1. Research Team Error Resiliece Aalysis of Multi-Hypothesis Motio Compesated Predictio for Video Codig Project Leader: Other Faculty: Post Doc(s): Graduate Studets: Idustrial Parter(s): C.-C. Jay Kuo,

More information

OPTIMAL PIECEWISE UNIFORM VECTOR QUANTIZATION OF THE MEMORYLESS LAPLACIAN SOURCE

OPTIMAL PIECEWISE UNIFORM VECTOR QUANTIZATION OF THE MEMORYLESS LAPLACIAN SOURCE Joural of ELECTRICAL EGIEERIG, VOL. 56, O. 7-8, 2005, 200 204 OPTIMAL PIECEWISE UIFORM VECTOR QUATIZATIO OF THE MEMORYLESS LAPLACIA SOURCE Zora H. Perić Veljo Lj. Staović Alesadra Z. Jovaović Srdja M.

More information

Vector Permutation Code Design Algorithm. Danilo SILVA and Weiler A. FINAMORE

Vector Permutation Code Design Algorithm. Danilo SILVA and Weiler A. FINAMORE Iteratioal Symposium o Iformatio Theory ad its Applicatios, ISITA2004 Parma, Italy, October 10 13, 2004 Vector Permutatio Code Desig Algorithm Dailo SILVA ad Weiler A. FINAMORE Cetro de Estudos em Telecomuicações

More information

Fundamentals. Relative data redundancy of the set. C.E., NCU, Taiwan Angela Chih-Wei Tang,

Fundamentals. Relative data redundancy of the set. C.E., NCU, Taiwan Angela Chih-Wei Tang, Image Compressio Agela Chih-Wei Tag ( 唐之瑋 ) Departmet of Commuicatio Egieerig Natioal Cetral Uiversity JhogLi, Taiwa 2012 Sprig Fudametals Compressio ratio C R / 2, 1 : origial, 2 1 : compressed Relative

More information

Topics Machine learning: lecture 2. Review: the learning problem. Hypotheses and estimation. Estimation criterion cont d. Estimation criterion

Topics Machine learning: lecture 2. Review: the learning problem. Hypotheses and estimation. Estimation criterion cont d. Estimation criterion .87 Machie learig: lecture Tommi S. Jaakkola MIT CSAIL tommi@csail.mit.edu Topics The learig problem hypothesis class, estimatio algorithm loss ad estimatio criterio samplig, empirical ad epected losses

More information

Sensitivity Analysis of Daubechies 4 Wavelet Coefficients for Reduction of Reconstructed Image Error

Sensitivity Analysis of Daubechies 4 Wavelet Coefficients for Reduction of Reconstructed Image Error Proceedigs of the 6th WSEAS Iteratioal Coferece o SIGNAL PROCESSING, Dallas, Texas, USA, March -4, 7 67 Sesitivity Aalysis of Daubechies 4 Wavelet Coefficiets for Reductio of Recostructed Image Error DEVINDER

More information

Structuring Element Representation of an Image and Its Applications

Structuring Element Representation of an Image and Its Applications Iteratioal Joural of Cotrol Structurig Automatio Elemet ad Represetatio Systems vol. of a o. Image 4 pp. ad 50955 Its Applicatios December 004 509 Structurig Elemet Represetatio of a Image ad Its Applicatios

More information

OFDM Precoder for Minimizing BER Upper Bound of MLD under Imperfect CSI

OFDM Precoder for Minimizing BER Upper Bound of MLD under Imperfect CSI MIMO-OFDM OFDM Precoder for Miimizig BER Upper Boud of MLD uder Imperfect CSI MCRG Joit Semiar Jue the th 008 Previously preseted at ICC 008 Beijig o May the st 008 Boosar Pitakdumrogkija Kazuhiko Fukawa

More information

An Efficient Lloyd-Max Quantizer for Matching Pursuit Decompositions

An Efficient Lloyd-Max Quantizer for Matching Pursuit Decompositions BRAZILVI ITERATIOAL TELECOMMUICATIOS SYMPOSIUM (ITS6), SEPTEMBER 3-6, 6, FORTALEZA-CE, BRAZIL A Efficiet Lloyd-Max Quatizer for Matchig Pursuit Decompositios Lisadro Lovisolo, Eduardo A B da Silva ad Paulo

More information

Discrete-Time Systems, LTI Systems, and Discrete-Time Convolution

Discrete-Time Systems, LTI Systems, and Discrete-Time Convolution EEL5: Discrete-Time Sigals ad Systems. Itroductio I this set of otes, we begi our mathematical treatmet of discrete-time s. As show i Figure, a discrete-time operates or trasforms some iput sequece x [

More information

Optimally Sparse SVMs

Optimally Sparse SVMs A. Proof of Lemma 3. We here prove a lower boud o the umber of support vectors to achieve geeralizatio bouds of the form which we cosider. Importatly, this result holds ot oly for liear classifiers, but

More information

APPLICATION OF CEPSTRUM ANALYSIS IN SPEECH CODING. Vahid Abolghasemi, Hossein Marvi

APPLICATION OF CEPSTRUM ANALYSIS IN SPEECH CODING. Vahid Abolghasemi, Hossein Marvi ICSV4 Cairs Australia 9- July, 7 APPLICATION OF CEPSTRUM ANALYSIS IN SPEECH CODING Vahid Abolghasemi, Hossei Marvi Faculty of Electrical & Robotic Egieerig, Shahrood Uiversity of Techology vahidabolghasemi@yahoo.com,

More information

NUMERICAL METHODS FOR SOLVING EQUATIONS

NUMERICAL METHODS FOR SOLVING EQUATIONS Mathematics Revisio Guides Numerical Methods for Solvig Equatios Page 1 of 11 M.K. HOME TUITION Mathematics Revisio Guides Level: GCSE Higher Tier NUMERICAL METHODS FOR SOLVING EQUATIONS Versio:. Date:

More information

w (1) ˆx w (1) x (1) /ρ and w (2) ˆx w (2) x (2) /ρ.

w (1) ˆx w (1) x (1) /ρ and w (2) ˆx w (2) x (2) /ρ. 2 5. Weighted umber of late jobs 5.1. Release dates ad due dates: maximimizig the weight of o-time jobs Oce we add release dates, miimizig the umber of late jobs becomes a sigificatly harder problem. For

More information

Module 5 EMBEDDED WAVELET CODING. Version 2 ECE IIT, Kharagpur

Module 5 EMBEDDED WAVELET CODING. Version 2 ECE IIT, Kharagpur Module 5 EMBEDDED WAVELET CODING Versio ECE IIT, Kharagpur Lesso 4 SPIHT algorithm Versio ECE IIT, Kharagpur Istructioal Objectives At the ed of this lesso, the studets should be able to:. State the limitatios

More information

Electricity consumption forecasting method based on MPSO-BP neural network model Youshan Zhang 1, 2,a, Liangdong Guo2, b,qi Li 3, c and Junhui Li2, d

Electricity consumption forecasting method based on MPSO-BP neural network model Youshan Zhang 1, 2,a, Liangdong Guo2, b,qi Li 3, c and Junhui Li2, d 4th Iteratioal Coferece o Electrical & Electroics Egieerig ad Computer Sciece (ICEEECS 2016) Electricity cosumptio forecastig method based o eural etwork model Yousha Zhag 1, 2,a, Liagdog Guo2, b,qi Li

More information

An Improved Proportionate Normalized Least Mean Square Algorithm with Orthogonal Correction Factors for Echo Cancellation

An Improved Proportionate Normalized Least Mean Square Algorithm with Orthogonal Correction Factors for Echo Cancellation 202 Iteratioal Coferece o Electroics Egieerig ad Iformatics (ICEEI 202) IPCSI vol. 49 (202) (202) IACSI Press, Sigapore DOI: 0.7763/IPCSI.202.V49.33 A Improved Proportioate Normalized Least Mea Square

More information

Goodness-Of-Fit For The Generalized Exponential Distribution. Abstract

Goodness-Of-Fit For The Generalized Exponential Distribution. Abstract Goodess-Of-Fit For The Geeralized Expoetial Distributio By Amal S. Hassa stitute of Statistical Studies & Research Cairo Uiversity Abstract Recetly a ew distributio called geeralized expoetial or expoetiated

More information

Extended Weighted Linear Prediction Using the Autocorrelation Snapshot

Extended Weighted Linear Prediction Using the Autocorrelation Snapshot Exteded Weighted Liear Predictio Usig the Autocorrelatio Sapshot - A Robust Speech Aalysis Method ad its Applicatio to Recogitio of Vocal Emotios Joui Pohjalaie ad Paavo Alku Departmet of Sigal Processig

More information

5. Fast NLMS-OCF Algorithm

5. Fast NLMS-OCF Algorithm 5. Fast LMS-OCF Algorithm The LMS-OCF algorithm preseted i Chapter, which relies o Gram-Schmidt orthogoalizatio, has a compleity O ( M ). The square-law depedece o computatioal requiremets o the umber

More information

Linear Associator Linear Layer

Linear Associator Linear Layer Hebbia Learig opic 6 Note: lecture otes by Michael Negevitsky (uiversity of asmaia) Bob Keller (Harvey Mudd College CA) ad Marti Haga (Uiversity of Colorado) are used Mai idea: learig based o associatio

More information

REGRESSION (Physics 1210 Notes, Partial Modified Appendix A)

REGRESSION (Physics 1210 Notes, Partial Modified Appendix A) REGRESSION (Physics 0 Notes, Partial Modified Appedix A) HOW TO PERFORM A LINEAR REGRESSION Cosider the followig data poits ad their graph (Table I ad Figure ): X Y 0 3 5 3 7 4 9 5 Table : Example Data

More information

Double Stage Shrinkage Estimator of Two Parameters. Generalized Exponential Distribution

Double Stage Shrinkage Estimator of Two Parameters. Generalized Exponential Distribution Iteratioal Mathematical Forum, Vol., 3, o. 3, 3-53 HIKARI Ltd, www.m-hikari.com http://dx.doi.org/.9/imf.3.335 Double Stage Shrikage Estimator of Two Parameters Geeralized Expoetial Distributio Alaa M.

More information

Parallel Vector Algorithms David A. Padua

Parallel Vector Algorithms David A. Padua Parallel Vector Algorithms 1 of 32 Itroductio Next, we study several algorithms where parallelism ca be easily expressed i terms of array operatios. We will use Fortra 90 to represet these algorithms.

More information

ECE 564/645 - Digital Communication Systems (Spring 2014) Final Exam Friday, May 2nd, 8:00-10:00am, Marston 220

ECE 564/645 - Digital Communication Systems (Spring 2014) Final Exam Friday, May 2nd, 8:00-10:00am, Marston 220 ECE 564/645 - Digital Commuicatio Systems (Sprig 014) Fial Exam Friday, May d, 8:00-10:00am, Marsto 0 Overview The exam cosists of four (or five) problems for 100 (or 10) poits. The poits for each part

More information

Mathematical Modeling of Optimum 3 Step Stress Accelerated Life Testing for Generalized Pareto Distribution

Mathematical Modeling of Optimum 3 Step Stress Accelerated Life Testing for Generalized Pareto Distribution America Joural of Theoretical ad Applied Statistics 05; 4(: 6-69 Published olie May 8, 05 (http://www.sciecepublishiggroup.com/j/ajtas doi: 0.648/j.ajtas.05040. ISSN: 6-8999 (Prit; ISSN: 6-9006 (Olie Mathematical

More information

(all terms are scalars).the minimization is clearer in sum notation:

(all terms are scalars).the minimization is clearer in sum notation: 7 Multiple liear regressio: with predictors) Depedet data set: y i i = 1, oe predictad, predictors x i,k i = 1,, k = 1, ' The forecast equatio is ŷ i = b + Use matrix otatio: k =1 b k x ik Y = y 1 y 1

More information

ADVANCED SOFTWARE ENGINEERING

ADVANCED SOFTWARE ENGINEERING ADVANCED SOFTWARE ENGINEERING COMP 3705 Exercise Usage-based Testig ad Reliability Versio 1.0-040406 Departmet of Computer Ssciece Sada Narayaappa, Aeliese Adrews Versio 1.1-050405 Departmet of Commuicatio

More information

Bayesian and E- Bayesian Method of Estimation of Parameter of Rayleigh Distribution- A Bayesian Approach under Linex Loss Function

Bayesian and E- Bayesian Method of Estimation of Parameter of Rayleigh Distribution- A Bayesian Approach under Linex Loss Function Iteratioal Joural of Statistics ad Systems ISSN 973-2675 Volume 12, Number 4 (217), pp. 791-796 Research Idia Publicatios http://www.ripublicatio.com Bayesia ad E- Bayesia Method of Estimatio of Parameter

More information

Complex Algorithms for Lattice Adaptive IIR Notch Filter

Complex Algorithms for Lattice Adaptive IIR Notch Filter 4th Iteratioal Coferece o Sigal Processig Systems (ICSPS ) IPCSIT vol. 58 () () IACSIT Press, Sigapore DOI:.7763/IPCSIT..V58. Complex Algorithms for Lattice Adaptive IIR Notch Filter Hog Liag +, Nig Jia

More information

Time-Domain Representations of LTI Systems

Time-Domain Representations of LTI Systems 2.1 Itroductio Objectives: 1. Impulse resposes of LTI systems 2. Liear costat-coefficiets differetial or differece equatios of LTI systems 3. Bloc diagram represetatios of LTI systems 4. State-variable

More information

1 Review of Probability & Statistics

1 Review of Probability & Statistics 1 Review of Probability & Statistics a. I a group of 000 people, it has bee reported that there are: 61 smokers 670 over 5 960 people who imbibe (drik alcohol) 86 smokers who imbibe 90 imbibers over 5

More information

ECE 901 Lecture 12: Complexity Regularization and the Squared Loss

ECE 901 Lecture 12: Complexity Regularization and the Squared Loss ECE 90 Lecture : Complexity Regularizatio ad the Squared Loss R. Nowak 5/7/009 I the previous lectures we made use of the Cheroff/Hoeffdig bouds for our aalysis of classifier errors. Hoeffdig s iequality

More information

Introduction to Signals and Systems, Part V: Lecture Summary

Introduction to Signals and Systems, Part V: Lecture Summary EEL33: Discrete-Time Sigals ad Systems Itroductio to Sigals ad Systems, Part V: Lecture Summary Itroductio to Sigals ad Systems, Part V: Lecture Summary So far we have oly looked at examples of o-recursive

More information

A new iterative algorithm for reconstructing a signal from its dyadic wavelet transform modulus maxima

A new iterative algorithm for reconstructing a signal from its dyadic wavelet transform modulus maxima ol 46 No 6 SCIENCE IN CHINA (Series F) December 3 A ew iterative algorithm for recostructig a sigal from its dyadic wavelet trasform modulus maxima ZHANG Zhuosheg ( u ), LIU Guizhog ( q) & LIU Feg ( )

More information

Report on Private Information Retrieval over Unsynchronized Databases

Report on Private Information Retrieval over Unsynchronized Databases Report o Private Iformatio Retrieval over Usychroized Databases Lembit Valgma Supervised by Vitaly Skachek May 25, 217 1 Problem Statemet There are may challeges cocerig olie privacy. Private iformatio

More information

1 Inferential Methods for Correlation and Regression Analysis

1 Inferential Methods for Correlation and Regression Analysis 1 Iferetial Methods for Correlatio ad Regressio Aalysis I the chapter o Correlatio ad Regressio Aalysis tools for describig bivariate cotiuous data were itroduced. The sample Pearso Correlatio Coefficiet

More information

Integer Linear Programming

Integer Linear Programming Iteger Liear Programmig Itroductio Iteger L P problem (P) Mi = s. t. a = b i =,, m = i i 0, iteger =,, c Eemple Mi z = 5 s. t. + 0 0, 0, iteger F(P) = feasible domai of P Itroductio Iteger L P problem

More information

Invariability of Remainder Based Reversible Watermarking

Invariability of Remainder Based Reversible Watermarking Joural of Network Itelligece c 16 ISSN 21-8105 (Olie) Taiwa Ubiquitous Iformatio Volume 1, Number 1, February 16 Ivariability of Remaider Based Reversible Watermarkig Shao-Wei Weg School of Iformatio Egieerig

More information

EE 505. Lecture 28. ADC Design SAR

EE 505. Lecture 28. ADC Design SAR EE 505 Lecture 28 ADC Desig SAR Review from Last Lecture Elimiatio of Iput S/H C LK X IN S/H Stage 1 r 1 Stage 2 r 2 Stage k r k Stage m r m 1 2 k m Pipelied Assembler (Shift Register

More information

( a) ( ) 1 ( ) 2 ( ) ( ) 3 3 ( ) =!

( a) ( ) 1 ( ) 2 ( ) ( ) 3 3 ( ) =! .8,.9: Taylor ad Maclauri Series.8. Although we were able to fid power series represetatios for a limited group of fuctios i the previous sectio, it is ot immediately obvious whether ay give fuctio has

More information

Summary: CORRELATION & LINEAR REGRESSION. GC. Students are advised to refer to lecture notes for the GC operations to obtain scatter diagram.

Summary: CORRELATION & LINEAR REGRESSION. GC. Students are advised to refer to lecture notes for the GC operations to obtain scatter diagram. Key Cocepts: 1) Sketchig of scatter diagram The scatter diagram of bivariate (i.e. cotaiig two variables) data ca be easily obtaied usig GC. Studets are advised to refer to lecture otes for the GC operatios

More information

Finite Block-Length Gains in Distributed Source Coding

Finite Block-Length Gains in Distributed Source Coding Decoder Fiite Block-Legth Gais i Distributed Source Codig Farhad Shirai EECS Departmet Uiversity of Michiga A Arbor,USA Email: fshirai@umichedu S Sadeep Pradha EECS Departmet Uiversity of Michiga A Arbor,USA

More information

Linear Regression Demystified

Linear Regression Demystified Liear Regressio Demystified Liear regressio is a importat subject i statistics. I elemetary statistics courses, formulae related to liear regressio are ofte stated without derivatio. This ote iteds to

More information

Cooperative Communication Fundamentals & Coding Techniques

Cooperative Communication Fundamentals & Coding Techniques 3 th ICACT Tutorial Cooperative commuicatio fudametals & codig techiques Cooperative Commuicatio Fudametals & Codig Techiques 0..4 Electroics ad Telecommuicatio Research Istitute Kiug Jug 3 th ICACT Tutorial

More information

The DOA Estimation of Multiple Signals based on Weighting MUSIC Algorithm

The DOA Estimation of Multiple Signals based on Weighting MUSIC Algorithm , pp.10-106 http://dx.doi.org/10.1457/astl.016.137.19 The DOA Estimatio of ultiple Sigals based o Weightig USIC Algorithm Chagga Shu a, Yumi Liu State Key Laboratory of IPOC, Beijig Uiversity of Posts

More information

Entropies & Information Theory

Entropies & Information Theory Etropies & Iformatio Theory LECTURE I Nilajaa Datta Uiversity of Cambridge,U.K. For more details: see lecture otes (Lecture 1- Lecture 5) o http://www.qi.damtp.cam.ac.uk/ode/223 Quatum Iformatio Theory

More information

Rank Modulation with Multiplicity

Rank Modulation with Multiplicity Rak Modulatio with Multiplicity Axiao (Adrew) Jiag Computer Sciece ad Eg. Dept. Texas A&M Uiversity College Statio, TX 778 ajiag@cse.tamu.edu Abstract Rak modulatio is a scheme that uses the relative order

More information

A statistical method to determine sample size to estimate characteristic value of soil parameters

A statistical method to determine sample size to estimate characteristic value of soil parameters A statistical method to determie sample size to estimate characteristic value of soil parameters Y. Hojo, B. Setiawa 2 ad M. Suzuki 3 Abstract Sample size is a importat factor to be cosidered i determiig

More information

You may work in pairs or purely individually for this assignment.

You may work in pairs or purely individually for this assignment. CS 04 Problem Solvig i Computer Sciece OOC Assigmet 6: Recurreces You may work i pairs or purely idividually for this assigmet. Prepare your aswers to the followig questios i a plai ASCII text file or

More information

Spectral Observer with Reduced Information Demand

Spectral Observer with Reduced Information Demand Spectral Observer with Reduced Iformatio Demad Gy. Orosz, L. Sujbert, G. Péceli Departmet of Measuremet ad Iformatio Systems, Budapest Uiversity of Techology ad Ecoomics Magyar tudósok krt. 2., -52 Budapest,

More information

There is no straightforward approach for choosing the warmup period l.

There is no straightforward approach for choosing the warmup period l. B. Maddah INDE 504 Discrete-Evet Simulatio Output Aalysis () Statistical Aalysis for Steady-State Parameters I a otermiatig simulatio, the iterest is i estimatig the log ru steady state measures of performace.

More information

Pixel Recurrent Neural Networks

Pixel Recurrent Neural Networks Pixel Recurret Neural Networks Aa ro va de Oord, Nal Kalchbreer, Koray Kavukcuoglu Google DeepMid August 2016 Preseter - Neha M Example problem (completig a image) Give the first half of the image, create

More information

ANOTHER WEIGHTED WEIBULL DISTRIBUTION FROM AZZALINI S FAMILY

ANOTHER WEIGHTED WEIBULL DISTRIBUTION FROM AZZALINI S FAMILY ANOTHER WEIGHTED WEIBULL DISTRIBUTION FROM AZZALINI S FAMILY Sulema Nasiru, MSc. Departmet of Statistics, Faculty of Mathematical Scieces, Uiversity for Developmet Studies, Navrogo, Upper East Regio, Ghaa,

More information

Transform-domain Temporal Prediction in Video Coding with Spatially Adaptive Spectral Correlations

Transform-domain Temporal Prediction in Video Coding with Spatially Adaptive Spectral Correlations Trasform-domai Temporal Predictio i Video Codig with Spatially Adaptive Spectral Correlatios Jigig Ha, Viay Melkote, ad Keeth Rose Departmet of Electrical ad Computer Egieerig, Uiversity of Califoria Sata

More information

A New Solution Method for the Finite-Horizon Discrete-Time EOQ Problem

A New Solution Method for the Finite-Horizon Discrete-Time EOQ Problem This is the Pre-Published Versio. A New Solutio Method for the Fiite-Horizo Discrete-Time EOQ Problem Chug-Lu Li Departmet of Logistics The Hog Kog Polytechic Uiversity Hug Hom, Kowloo, Hog Kog Phoe: +852-2766-7410

More information

DERIVING THE 12-LEAD ECG FROM EASI ELECTRODES VIA NONLINEAR REGRESSION

DERIVING THE 12-LEAD ECG FROM EASI ELECTRODES VIA NONLINEAR REGRESSION DERIVING THE 1-LEAD ECG FROM EASI ELECTRODES VIA NONLINEAR REGRESSION 1 PIROON KAEWFOONGRUNGSI, DARANEE HORMDEE 1, Departmet of Computer Egieerig, Faculty of Egieerig, Kho Kae Uiversity 4, Thailad E-mail:

More information

STUDY OF ATTRACTOR VARIATION IN THE RECONSTRUCTED PHASE SPACE OF SPEECH SIGNALS

STUDY OF ATTRACTOR VARIATION IN THE RECONSTRUCTED PHASE SPACE OF SPEECH SIGNALS STUDY OF ATTRACTOR VARIATION IN THE RECONSTRUCTED PHASE SPACE OF SPEECH SIGNALS Jiji Ye Departmet of Electrical ad Computer Egieerig Milwaukee, WI USA jiji.ye@mu.edu Michael T. Johso Departmet of Electrical

More information

Information Theory and Coding

Information Theory and Coding Sol. Iformatio Theory ad Codig. The capacity of a bad-limited additive white Gaussia (AWGN) chael is give by C = Wlog 2 ( + σ 2 W ) bits per secod(bps), where W is the chael badwidth, is the average power

More information

6.867 Machine learning, lecture 7 (Jaakkola) 1

6.867 Machine learning, lecture 7 (Jaakkola) 1 6.867 Machie learig, lecture 7 (Jaakkola) 1 Lecture topics: Kerel form of liear regressio Kerels, examples, costructio, properties Liear regressio ad kerels Cosider a slightly simpler model where we omit

More information

Probability of error for LDPC OC with one co-channel Interferer over i.i.d Rayleigh Fading

Probability of error for LDPC OC with one co-channel Interferer over i.i.d Rayleigh Fading IOSR Joural of Electroics ad Commuicatio Egieerig (IOSR-JECE) e-issn: 2278-2834,p- ISSN: 2278-8735.Volume 9, Issue 4, Ver. III (Jul - Aug. 24), PP 59-63 Probability of error for LDPC OC with oe co-chael

More information

62. Power series Definition 16. (Power series) Given a sequence {c n }, the series. c n x n = c 0 + c 1 x + c 2 x 2 + c 3 x 3 +

62. Power series Definition 16. (Power series) Given a sequence {c n }, the series. c n x n = c 0 + c 1 x + c 2 x 2 + c 3 x 3 + 62. Power series Defiitio 16. (Power series) Give a sequece {c }, the series c x = c 0 + c 1 x + c 2 x 2 + c 3 x 3 + is called a power series i the variable x. The umbers c are called the coefficiets of

More information

ANALYSIS OF EXPERIMENTAL ERRORS

ANALYSIS OF EXPERIMENTAL ERRORS ANALYSIS OF EXPERIMENTAL ERRORS All physical measuremets ecoutered i the verificatio of physics theories ad cocepts are subject to ucertaities that deped o the measurig istrumets used ad the coditios uder

More information

Chapter 6: BINOMIAL PROBABILITIES

Chapter 6: BINOMIAL PROBABILITIES Charles Bocelet, Probability, Statistics, ad Radom Sigals," Oxford Uiversity Press, 016. ISBN: 978-0-19-00051-0 Chater 6: BINOMIAL PROBABILITIES Sectios 6.1 Basics of the Biomial Distributio 6. Comutig

More information

ECE-S352 Introduction to Digital Signal Processing Lecture 3A Direct Solution of Difference Equations

ECE-S352 Introduction to Digital Signal Processing Lecture 3A Direct Solution of Difference Equations ECE-S352 Itroductio to Digital Sigal Processig Lecture 3A Direct Solutio of Differece Equatios Discrete Time Systems Described by Differece Equatios Uit impulse (sample) respose h() of a DT system allows

More information

ECONOMIC OPERATION OF POWER SYSTEMS

ECONOMIC OPERATION OF POWER SYSTEMS ECOOMC OEATO OF OWE SYSTEMS TOUCTO Oe of the earliest applicatios of o-lie cetralized cotrol was to provide a cetral facility, to operate ecoomically, several geeratig plats supplyig the loads of the system.

More information

Asymptotic Coupling and Its Applications in Information Theory

Asymptotic Coupling and Its Applications in Information Theory Asymptotic Couplig ad Its Applicatios i Iformatio Theory Vicet Y. F. Ta Joit Work with Lei Yu Departmet of Electrical ad Computer Egieerig, Departmet of Mathematics, Natioal Uiversity of Sigapore IMS-APRM

More information

Filter banks. Separately, the lowpass and highpass filters are not invertible. removes the highest frequency 1/ 2and

Filter banks. Separately, the lowpass and highpass filters are not invertible. removes the highest frequency 1/ 2and Filter bas Separately, the lowpass ad highpass filters are ot ivertible T removes the highest frequecy / ad removes the lowest frequecy Together these filters separate the sigal ito low-frequecy ad high-frequecy

More information

Linear Regression Models

Linear Regression Models Liear Regressio Models Dr. Joh Mellor-Crummey Departmet of Computer Sciece Rice Uiversity johmc@cs.rice.edu COMP 528 Lecture 9 15 February 2005 Goals for Today Uderstad how to Use scatter diagrams to ispect

More information

ELEC1200: A System View of Communications: from Signals to Packets Lecture 3

ELEC1200: A System View of Communications: from Signals to Packets Lecture 3 ELEC2: A System View of Commuicatios: from Sigals to Packets Lecture 3 Commuicatio chaels Discrete time Chael Modelig the chael Liear Time Ivariat Systems Step Respose Respose to sigle bit Respose to geeral

More information

Resampling Methods. X (1/2), i.e., Pr (X i m) = 1/2. We order the data: X (1) X (2) X (n). Define the sample median: ( n.

Resampling Methods. X (1/2), i.e., Pr (X i m) = 1/2. We order the data: X (1) X (2) X (n). Define the sample median: ( n. Jauary 1, 2019 Resamplig Methods Motivatio We have so may estimators with the property θ θ d N 0, σ 2 We ca also write θ a N θ, σ 2 /, where a meas approximately distributed as Oce we have a cosistet estimator

More information

Analysis of Algorithms. Introduction. Contents

Analysis of Algorithms. Introduction. Contents Itroductio The focus of this module is mathematical aspects of algorithms. Our mai focus is aalysis of algorithms, which meas evaluatig efficiecy of algorithms by aalytical ad mathematical methods. We

More information

Sliding Mode Indicator Model s at Thermal Control System for Furnace

Sliding Mode Indicator Model s at Thermal Control System for Furnace Proceedigs of the 7th World Cogress The Iteratioal Federatio of Automatic Cotrol Slidig Mode Idicator Model s at Thermal Cotrol System for Furace Varda Mkrttchia*, Sargis Simoya**, Aa Khachaturova***,

More information

Comparison Study of Series Approximation. and Convergence between Chebyshev. and Legendre Series

Comparison Study of Series Approximation. and Convergence between Chebyshev. and Legendre Series Applied Mathematical Scieces, Vol. 7, 03, o. 6, 3-337 HIKARI Ltd, www.m-hikari.com http://d.doi.org/0.988/ams.03.3430 Compariso Study of Series Approimatio ad Covergece betwee Chebyshev ad Legedre Series

More information

DIRECT LINEAR CONVERSION OF LSP PARAMETERS FOR PERCEPTUAL CONTROL IN SPEECH AND AUDIO CODING

DIRECT LINEAR CONVERSION OF LSP PARAMETERS FOR PERCEPTUAL CONTROL IN SPEECH AND AUDIO CODING DIRECT LINEAR CONVERSION OF LSP PARAMETERS FOR PERCEPTUAL CONTROL IN SPEECH AND AUDIO CODING R Sugiura 1, Y Kamamoto 2, N Harada 2, H Kameoka 2, T Moriya 2 1 Graduate School of Iformatio Sciece ad Techology,

More information

UC Berkeley CS 170: Efficient Algorithms and Intractable Problems Handout 17 Lecturer: David Wagner April 3, Notes 17 for CS 170

UC Berkeley CS 170: Efficient Algorithms and Intractable Problems Handout 17 Lecturer: David Wagner April 3, Notes 17 for CS 170 UC Berkeley CS 170: Efficiet Algorithms ad Itractable Problems Hadout 17 Lecturer: David Wager April 3, 2003 Notes 17 for CS 170 1 The Lempel-Ziv algorithm There is a sese i which the Huffma codig was

More information

1.0 Probability of Error for non-coherent BFSK

1.0 Probability of Error for non-coherent BFSK Probability of Error, Digital Sigalig o a Fadig Chael Ad Equalizatio Schemes for ISI Wireless Commuicatios echologies Sprig 5 Lectures & R Departmet of Electrical Egieerig, Rutgers Uiversity, Piscataway,

More information

Dual Frame Motion Compensation with Optimal Long-term Reference Frame Selection and Bit Allocation

Dual Frame Motion Compensation with Optimal Long-term Reference Frame Selection and Bit Allocation AER IDENTIFICATION NUMBER:33 Dual Frame Motio Compesatio with Optimal Log-term Referece Frame Selectio ad Bit Allocatio Da Liu, Debi Zhao, Xiagyag i, ad We Gao, Fellow, IEEE Abstract I dual frame motio

More information

A Unified Approach on Fast Training of Feedforward and Recurrent Networks Using EM Algorithm

A Unified Approach on Fast Training of Feedforward and Recurrent Networks Using EM Algorithm 2270 IEEE TRASACTIOS O SIGAL PROCESSIG, VOL. 46, O. 8, AUGUST 1998 [12] Q. T. Zhag, K. M. Wog, P. C. Yip, ad J. P. Reilly, Statistical aalysis of the performace of iformatio criteria i the detectio of

More information

Control Charts for Mean for Non-Normally Correlated Data

Control Charts for Mean for Non-Normally Correlated Data Joural of Moder Applied Statistical Methods Volume 16 Issue 1 Article 5 5-1-017 Cotrol Charts for Mea for No-Normally Correlated Data J. R. Sigh Vikram Uiversity, Ujjai, Idia Ab Latif Dar School of Studies

More information

Compound conditional source coding, Slepian-Wolf list decoding, and applications to media coding

Compound conditional source coding, Slepian-Wolf list decoding, and applications to media coding Compoud coditioal source codig, Slepia-Wolf list decodig, ad applicatios to media codig Stark C. Draper Mitsubishi Electric Research Laboratories Cambridge, MA 02139 USA draper@merl.com PSfrag replacemets

More information

Disjoint set (Union-Find)

Disjoint set (Union-Find) CS124 Lecture 7 Fall 2018 Disjoit set (Uio-Fid) For Kruskal s algorithm for the miimum spaig tree problem, we foud that we eeded a data structure for maitaiig a collectio of disjoit sets. That is, we eed

More information

TR/46 OCTOBER THE ZEROS OF PARTIAL SUMS OF A MACLAURIN EXPANSION A. TALBOT

TR/46 OCTOBER THE ZEROS OF PARTIAL SUMS OF A MACLAURIN EXPANSION A. TALBOT TR/46 OCTOBER 974 THE ZEROS OF PARTIAL SUMS OF A MACLAURIN EXPANSION by A. TALBOT .. Itroductio. A problem i approximatio theory o which I have recetly worked [] required for its solutio a proof that the

More information

Characterization and Simulation of HF Propagation in a realistic disturbed and horizontally inhomogeneous Ionosphere

Characterization and Simulation of HF Propagation in a realistic disturbed and horizontally inhomogeneous Ionosphere Characterizatio ad Simulatio of HF Propagatio i a realistic disturbed ad horizotally ihomogeeous Ioosphere Vadim E. Gherm ad Nikolay N.Zerov Departmet of Radiophysics Uiversity of St.Petersburg St.Petersburg

More information

OPTIMAL ALGORITHMS -- SUPPLEMENTAL NOTES

OPTIMAL ALGORITHMS -- SUPPLEMENTAL NOTES OPTIMAL ALGORITHMS -- SUPPLEMENTAL NOTES Peter M. Maurer Why Hashig is θ(). As i biary search, hashig assumes that keys are stored i a array which is idexed by a iteger. However, hashig attempts to bypass

More information

SRC Technical Note June 17, Tight Thresholds for The Pure Literal Rule. Michael Mitzenmacher. d i g i t a l

SRC Technical Note June 17, Tight Thresholds for The Pure Literal Rule. Michael Mitzenmacher. d i g i t a l SRC Techical Note 1997-011 Jue 17, 1997 Tight Thresholds for The Pure Literal Rule Michael Mitzemacher d i g i t a l Systems Research Ceter 130 Lytto Aveue Palo Alto, Califoria 94301 http://www.research.digital.com/src/

More information

Non-Linear Maximum Likelihood Feature Transformation For Speech Recognition

Non-Linear Maximum Likelihood Feature Transformation For Speech Recognition No-Liear Maximum Likelihood Feature Trasformatio For Speech Recogitio Mohamed Kamal Omar, Mark Hasegawa-Johso Departmet of Electrical Ad Computer Egieerig, Uiversity of Illiois at Urbaa-Champaig, Urbaa,

More information

Computational Tutorial of Steepest Descent Method and Its Implementation in Digital Image Processing

Computational Tutorial of Steepest Descent Method and Its Implementation in Digital Image Processing Computatioal utorial of Steepest Descet Method ad Its Implemetatio i Digital Image Processig Vorapoj Pataavijit Departmet of Electrical ad Electroic Egieerig, Faculty of Egieerig Assumptio Uiversity, Bagkok,

More information

7. Modern Techniques. Data Encryption Standard (DES)

7. Modern Techniques. Data Encryption Standard (DES) 7. Moder Techiques. Data Ecryptio Stadard (DES) The objective of this chapter is to illustrate the priciples of moder covetioal ecryptio. For this purpose, we focus o the most widely used covetioal ecryptio

More information

Machine Learning Theory Tübingen University, WS 2016/2017 Lecture 11

Machine Learning Theory Tübingen University, WS 2016/2017 Lecture 11 Machie Learig Theory Tübige Uiversity, WS 06/07 Lecture Tolstikhi Ilya Abstract We will itroduce the otio of reproducig kerels ad associated Reproducig Kerel Hilbert Spaces (RKHS). We will cosider couple

More information

The Phi Power Series

The Phi Power Series The Phi Power Series I did this work i about 0 years while poderig the relatioship betwee the golde mea ad the Madelbrot set. I have fially decided to make it available from my blog at http://semresearch.wordpress.com/.

More information

Estimating the Change Point of Bivariate Binomial Processes Experiencing Step Changes in Their Mean

Estimating the Change Point of Bivariate Binomial Processes Experiencing Step Changes in Their Mean Proceedigs of the 202 Iteratioal Coferece o Idustrial Egieerig ad Operatios Maagemet Istabul, Turey, July 3 6, 202 Estimatig the Chage Poit of Bivariate Biomial Processes Experiecig Step Chages i Their

More information

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur Module 3 LOSSY IMAGE COMPRESSION SYSEMS Versio ECE II, Kharagpur Lesso 8 rasform Codig & K-L rasforms Versio ECE II, Kharagpur Istructioal Oectives At the ed of this lesso, the studets should e ale to:.

More information

A New Class of Ternary Zero Correlation Zone Sequence Sets Based on Mutually Orthogonal Complementary Sets

A New Class of Ternary Zero Correlation Zone Sequence Sets Based on Mutually Orthogonal Complementary Sets IOSR Joural of Electroics ad Commuicatio Egieerig (IOSR-JECE) e-issn: 2278-2834,p- ISSN: 2278-8735.Volume 0, Issue 3, Ver. I (May - Ju.205), PP 08-3 www.iosrjourals.org A New Class of Terary Zero Correlatio

More information