PATTERN'recognition' and MACHINE'learning CHAPTER'5:'NEURAL'networks Include(Nonlinearity(G(X)(In(Output( With(Respect(To(Input

Total Page:16

File Type:pdf, Size:1020Kb

PATTERN'recognition' and MACHINE'learning CHAPTER'5:'NEURAL'networks Include(Nonlinearity(G(X)(In(Output( With(Respect(To(Input PATTERN'RECOGNITION' AND MACHINE'LEARNING CHAPTER'5:'NEURAL'NETWORKS Include(Nonlinearity(g(x)(in(Output( with(Respect(to(Input Yk(X)(=(g((! wki(xi( +(wk0((),(k(=(1,(…,(K Y1(X)$…$$$$Yk(X) g($) g($) Output w_11$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$w_1d$$$$$$$$$$$$$$$$$$w_0 Input x_1$$$x_2$$$$$x_3$$$$x_4$$$$$…$$$$$$$$x_d$$$$$$$$$$$$$$$$x_0 Using$sigmoid$nonlinearity$and$normal$class$conditional$probability,$the$ output$oF$a$NN$discriminant$is$interpreted$as$posteriori$probabilities$p(x|Ck) Mapping'Arbitrary'Boolean'Function Input&vector:' length'd,'all'components'0'or'1 Output: 1:'if'the'given'input'is'class'A,'0:'if'input'from'class'B Total'2^d'inputs;'say'K'are'in'class'A,'2^d'C K'in'B 2&layers&of&FF&NN Input'size'2^d;'Hidden'size'K;'Output'size'1;'hardlim threshold'funct. Weights Input'C>'Hidden:'1'if'given'input'is'in'A'and'has'1at'the'node;'C1'otherw Hidden'C>'Out:'all'1;'Bias/hidden:'1Cb:if'node'K'has'b'ones;' Prove!:'this'NN'gives'1'if'input'from'A,'0'from'B. Mapping'Arbitrary'Function' with'34layer'FFNN Single'neuron'threshold 4>'half4space 2'layer'NN 4>'convex'region Output'bias:'4M(hidden'u.)'gives'logical'AND 3'layer'NN 4>'any'region'!!! Subdivide'input'into'approx.'hypercubes A'cluster'of'd'1st'hidden'nodes'maps'one'cube Bias'41'means'logical'OR'at'output Can'produce'any'combination'of'input'cubes' 3"Layer(Neural(Network((1(hidden(layer) Kolmogorov(Approximation(Theorem( (1957) Discovered(independently(of(NNs Related(to(Hilbert’s 23(unsolved(problems/1900 #13:(Can(a(function(of(several(variaBles(Be(represented(as(a( comBination(of(functions(of(a(few(variaBle.(Arnold:(yes(N 3( variaBle(with(2! Kolmogorov(AN: Any(multivariaBle(continuous(function(can(Be(expressed(as( superposition(of(functions(of(one(variaBle((a(small(numBer( of(components) Limitations: not(constructive,(too(complicated Example:)3+layer,)feedforward)ANN) with)supervised)learning Motivation)for)3+layer:)Kolmogorov)Representation)Theorem y y8 9 9 8 w94 w w w87 w 95 w w96 86 84 85 w97 4 5 6 7 w61 w w w53 w63 w w41 w51 w 52 62 w 73 42 w72 43 w71 1 2 3 x2 x3 x1 Example:)3+layer,)FF)ANN(cont’d) Assume:)transfer)function)is)just)summation v v v 4 5 6 v7 4 5 6 7 w61 w w62 w53 w63 w w41 w51 w 52 w 73 42 w72 43 w71 1 2 3 x x x1 2 3 &v # &w w w # 4 41 42 43 &x # Input0to0 $v ! $w w w ! 1 $ 5 ! = $ 42 52 53 !$x ! hidden0 $ 2 ! $v6 ! $w 43 w 62 w 63 ! weights $ ! $ !%$x3 "! %v7 " %w 44 w 72 w 73 " Example:)3+layer,)FF)ANN(cont’d) hidden)to)output)weights &v4 # y w w w w $v ! & 8 # & 84 85 86 87 #$ 5 ! $ ! = $ ! %y9 " %w 94 w 95 w 96 w 97 "$v6 ! $ ! %v7 " y y8 9 9 8 w 94 w w95 w w86 87 w84 w85 96 v v v 4 5 6 v7 Example:&34layer,&FF&ANN(cont’d) Final&matrix&representation&for&linear&system V = WBAX Y = WcbV = WCBWBAX y9 &w 41 w 42 w 43 # y8 $ !&x1 # &y # &w w w w # w w w 8 = 84 85 86 87 $ 42 52 53 !$x ! $ ! $ ! $ 2 ! %y9 " %w 94 w 95 w 96 w 97 "$w 43 w 62 w 63 ! $ !%$x3 "!9 8 Layer1C %w 44 w 72 w 73 " w94 w w w87 w 95 w w96 86 84 85 w97 Layer1B 4 5 6 7 w61 w w w53 w63 w w41 w51 w 52 62 w 73 42 w72 43 w71 Layer1A 1 2 3 x1 x2 x3 Transfer(Functions May(be(a(threshold(function(that(passes(information( ONLY(IF(the(output(exceeds(the(threshold Can(be(a(continuous(function(of(the(input The(output(is(usually(passed(to(the(output(path(of( the(node Example(of(transfer(function:(sigmoid Examples)of)Approximation) (with)3)hidden)nodes)) Approximation+by+gradient+descent Often+not+practical+to+directly+evaluate Use+approximation+of+the+error+function+by+iteration: dE/dw =+0+ Gradient+descent+idea:+! if+we+are+at+a+given+location+ (given+w+parameters+to+be+optimized],+then+change+w+in+ a+way+where+the+gradient+of+the+error+is+the+max W(t+1)+=+W(t)+– !dE/dW Continue+the+iteration+until+Converges+to+the+minimum Not+always+converges! Illustration+of+the+Error+Surface Learning(with( Error(Backpropagation((BP) Learning:( determine(the(weights(of(the(NN Assume: Structure(is(given Transfer(functions(are(given Input(A output(pair(are(given( Supervised(learning(based(on(examples! See:(derivation(of(backpropagation Backpropagation,Learning Paul,Werbos,(1974) PhD,work,Princeton Roots,of,backpropagation Rumelhart,,McClelanD,(1986) PDP,group,at,CMU Popularization,of,the,iDea http://scsnl.stanford.eDu/conferences/NSF_Brain_Network_Dynamics_Jan2007 http://www.archive.org/search.php?query=2007+brain+network+Dynamics Supervised*Learning*Scheme Standard'Backpropagation'– delta'rule Gradient'of'the'sum'squared'error & 'F 'F 'F # ( ) F(w) = ($ , ,..., ! w $ ! % 'w1 'w2 'wQ " Backpropagation'delta9rule N N %F 1 %Fk 1 k k = lim = lim & li z(l"1) j N $# ! N $# ! %wp N k =1 %wlij N k =1 Weight'change'algorithm'iteratively'from'top'layer' N backward new old 1 k k wlij = wlij !% lim & li z(l!1) j N #" $ N k =1 or new old k k wlij = wlij !%& li z(l!1) j Generalized ! -rule 2 1 N where: FYxYxw= * ()! (,) Fw()= lim Fk (); w k kk N !" # N k =1 N batch size #Fw() 1N #Fw() = lim k N !" $ ##wNwppk =1 !!!FFIk kk kkei = k " = zFIli= () li !!!wIwlij li lij l-th layer $$FF$ "#kk ith - node = kk% wz= z kk&'* lij(1) l!! q (1) l j $$Iwli lij()q $ I li kk Iwzli= " liq(1) l! q q ##FF kk==!!kkz ; k ##wIli(1) l" j li lij li ... w k lij k z(1)1l! k z z(1)maxlq! (1)2l! ... k ... z(1)lj! k We have to calculate !li! =1 1 m $F $$zFk "#**2 l: output layer: k *li= k '(%&%& ykk) z= ))2 y kk z 2 k kk3*+*+p lp i li $F 2 $$zli Izli $ li '(,-,- ! k ==k ./p=1 li k 4 $$$Fzk F l: not output layer: k* li== k F0 Ik $Ili kk k( li ) 2 $$zIli li $ z li k 52 !"µl+1 #F #I k * (1)lp+ FI$ k ==%&) kk( li ) '(%&p=1 ##Iz(1)l+ p li !"µl+1 #F # $% k kk& =+='(kk* )*wzFI(1)l+ p lr( li ) // -.'(pr=1 ##Iz(1)l+ p li +, "#µµll++11$F k %%kk k ==&'k **()wFI(1)l+++ pi( li) ! (1)(1) l p wFI l pi li ** ()&'pp==11$I(1)lp+ N %F 1 kk k = !!liz(1) l" j where li is determined iteratively (see above) lim & %wNp N #$ k =1 N algorithm: weight change is new old1 k k wwlij= lij# !"% li z(1) l# j !F N k =1 proportional to (gradient): !wlij $ 0< ! ~small ( learning rate) Convergence)of)Backpropagation 1. Standard)backpropagation)reduces)error)F 9 BUT:)no)guarantee)of)convergence)to)global) minimum 9 Possible)problems: 9 Local)minimum 9 Very)slow)decrease 2. Theorem)on)the)approximation)by)39layer)NN Any)square)integrable)function)can)be)approximated) with)arbitrary)accuracy)by)a)39layer)backpropagation) ANN. 9 BUT:)no)guarantee)that)BP)(delta9rule)or)other)) gives)the)optimum)approximation Theorem'on'opt.'approx.'by'backpropagation D.W.Ruck,)S.K.Rogers,…)IEEE)Trans.)Neur.)Netw.,Vol.)1.pp.296A298,) 1990. Two'classes: p(x):=p(x|w1)P(w1)'+'p(x|w2)P(w2) p(x) @probability'distribution'of'feature'vectors'x''''''''''''''''''''p(x|wi)@ conditional'probability'density'f.'of'class'wi P(wi) @’a@priori’' probability'of'class'wi,'i=1,2.'''''''''''''P(wi|x) @’a@posteriori’' probability'of'class'x'to'belong'to'class'I Bayes'Rule: p(x|wi)P(wi)'='P(wi|x)p(x) Bayes'Discriminant:'''''P(w1|x)'– P(w2|x)'>'0 !select'class'1 THEOREM'(approximation'by'BPNN) An'optimally'selected'BP'NN'approximates'the'Bayesian'(maximum)' discriminant'function. NOTE:'the'actual'approximation'depends'on'the'structure'of'the' network,'class'conditional'probabilities'etc. Local&quadratic&approximation Taylor&expansion&of&error&function&w.r.t.&weights 1st and&2nd order:&gradient&and&Hessian&(H): Gradient&of&the&error: Modifications+of+Standard+Backpropagation 1. Optimum+choice+of+learning+rate: Initialization+of+weights Adaptive+learning+rate Randomization 2. Adding+a+momentum+term w k+1 = w k +#(1! µ)" k xk + µ(w k ! w k-1) 3. Regularization+term+to+SSE e.g.,+sum+of+weights+! pruning Forgetting+rate+! pruning * 2 I = !(yi " yi ) + ε'!| w ij |, i i, j Basic&NN&Architectures Feed$forward*NN Directed&graph&in&which&a&path&never&visits&the&same&node&twice Relatively&simple&behavior Example:&MLP&for&classification,&pattern&recognition Feedback*or*Recurrent*NNs Contains&loops&of&directed&edges&going&forward&and&also&backward Complicated&oscillations&might&occur Example:&Hopfield&NN,&Elman&NN&for&speech&recogn. Random*NNs More&realistic,&very&complex.
Recommended publications
  • PATTERN RECOGNITION LETTERS an Official Publication of the International Association for Pattern Recognition
    PATTERN RECOGNITION LETTERS An official publication of the International Association for Pattern Recognition AUTHOR INFORMATION PACK TABLE OF CONTENTS XXX . • Description p.1 • Audience p.2 • Impact Factor p.2 • Abstracting and Indexing p.2 • Editorial Board p.2 • Guide for Authors p.5 ISSN: 0167-8655 DESCRIPTION . Pattern Recognition Letters aims at rapid publication of concise articles of a broad interest in pattern recognition. Subject areas include all the current fields of interest represented by the Technical Committees of the International Association of Pattern Recognition, and other developing themes involving learning and recognition. Examples include: • Statistical, structural, syntactic pattern recognition; • Neural networks, machine learning, data mining; • Discrete geometry, algebraic, graph-based techniques for pattern recognition; • Signal analysis, image coding and processing, shape and texture analysis; • Computer vision, robotics, remote sensing; • Document processing, text and graphics recognition, digital libraries; • Speech recognition, music analysis, multimedia systems; • Natural language analysis, information retrieval; • Biometrics, biomedical pattern analysis and information systems; • Special hardware architectures, software packages for pattern recognition. We invite contributions as research reports or commentaries. Research reports should be concise summaries of methodological inventions and findings, with strong potential of wide applications. Alternatively, they can describe significant and novel applications
    [Show full text]
  • Library of Congress Classification
    R MEDICINE (GENERAL) R Medicine (General) Periodicals. Societies. Serials 5 International periodicals and serials 10 Medical societies Including aims, scope, utility, etc. International societies 10.5.A3 General works 10.5.A5-Z Individual societies America English United States. Canada 11 Periodicals. Serials 15 Societies British West Indies. Belize. Guyana 18 Periodicals. Serials 20 Societies Spanish and Portuguese Latin America 21 Periodicals. Serials 25 Societies 27.A-Z Other, A-Z 27.F7 French Europe English 31 Periodicals. Serials 35 Societies Dutch 37 Periodicals. Serials 39 Societies French 41 Periodicals. Serials 45 Societies German 51 Periodicals. Serials 55 Societies Italian 61 Periodicals. Serials 65 Societies Spanish and Portuguese 71 Periodicals. Serials 75 Societies Scandinavian 81 Periodicals. Serials 85 Societies Slavic 91 Periodicals. Serials 95 Societies 96.A-Z Other European languages, A-Z 96.H8 Hungarian Asia 97 English 97.5.A-Z Other European languages, A-Z 97.7.A-Z Other languages, A-Z Africa 98 English 98.5.A-Z Other European languages, A-Z 98.7.A-Z Other languages, A-Z 1 R MEDICINE (GENERAL) R Periodicals. Societies. Serials -- Continued Australasia and Pacific islands 99 English 99.5.A-Z Other European languages, A-Z 99.7.A-Z Other languages, A-Z Indexes see Z6658+ (101) Yearbooks see R5+ 104 Calendars. Almanacs Cf. AY81.M4 American popular medical almanacs 106 Congresses 108 Medical laboratories, institutes, etc. Class here papers and proceedings For works about these organizations see R860+ Collected works (nonserial) Cf. R126+ Ancient Greek and Latin works 111 Several authors 114 Individual authors Communication in medicine Cf.
    [Show full text]
  • New Summer Season
    New Summer Season Furniture Catalog — Live life outdoors® — Nomah® Nomah® 3 Seat Sofa W90" x D36" x H30" $4,699 (Slub Linens fabric) Nomah® Lounge Chair W40" x D36" x H30" $2,749 (Slub Linens fabric) Nomah® Coffee Table W40" x D40" x H12" $599 Choose your own cover from our outdoor fabric range. 5 Bronte The staple table for summer entertaining a b Find our furniture range online at ecooutdoorusa.com a. Bronte Dining Table W87" x D40" x H30" W107" x D40" x H30" W166" x D40" x H30" $2,499 | $2,899 | $4,999 b. Bronte Bench Seat Bronte Stool W69" x D18" x H19" W500 x D500 x H460 W87" x D18" x H19" $399 $999 | $1,199 6 ® Barwon Sophisticated, European-inspired design a. Barwon® Dining Chair W20" x D25" x H35" $499 b. Barwon® Dining Armchair W23" x D25" x H35" $549 c. Barwon® Bar Stool W20" x D25" x H45" $649 ® d. Barwon Side Table a W18" x D16" x H16" $199 Find our furniture range online at ecooutdoorusa.com b c d 9 Porto & Hux Modern & slimline design a. Porto Dining Table W71" x D36" x H30" $1,599 b. Porto Extending Dining Table W87-111" x D40" x H30" a $2,599 c. Porto Dining Table W36" x D36" x H30" $899 d. Porto Bench W63" x D18" x H19" W78" x D18" x H19" $799 | $999 e. Hux Dining Chair W19" x D21" x H34" b $319 Find our furniture range online at ecooutdoorusa.com c d e 10 The Hayfield combines unique curved arms Hayfield and refined componentry for a timeless look.
    [Show full text]
  • A Wavenet for Speech Denoising
    A Wavenet for Speech Denoising Dario Rethage∗ Jordi Pons∗ Xavier Serra [email protected] [email protected] [email protected] Music Technology Group Music Technology Group Music Technology Group Universitat Pompeu Fabra Universitat Pompeu Fabra Universitat Pompeu Fabra Abstract Currently, most speech processing techniques use magnitude spectrograms as front- end and are therefore by default discarding part of the signal: the phase. In order to overcome this limitation, we propose an end-to-end learning method for speech denoising based on Wavenet. The proposed model adaptation retains Wavenet’s powerful acoustic modeling capabilities, while significantly reducing its time- complexity by eliminating its autoregressive nature. Specifically, the model makes use of non-causal, dilated convolutions and predicts target fields instead of a single target sample. The discriminative adaptation of the model we propose, learns in a supervised fashion via minimizing a regression loss. These modifications make the model highly parallelizable during both training and inference. Both computational and perceptual evaluations indicate that the proposed method is preferred to Wiener filtering, a common method based on processing the magnitude spectrogram. 1 Introduction Over the last several decades, machine learning has produced solutions to complex problems that were previously unattainable with signal processing techniques [4, 12, 38]. Speech recognition is one such problem where machine learning has had a very strong impact. However, until today it has been standard practice not to work directly in the time-domain, but rather to explicitly use time-frequency representations as input [1, 34, 35] – for reducing the high-dimensionality of raw waveforms. Similarly, most techniques for speech denoising use magnitude spectrograms as front-end [13, 17, 21, 34, 36].
    [Show full text]
  • Heater Element Specifications Bulletin Number 592
    Technical Data Heater Element Specifications Bulletin Number 592 Topic Page Description 2 Heater Element Selection Procedure 2 Index to Heater Element Selection Tables 5 Heater Element Selection Tables 6 Additional Resources These documents contain additional information concerning related products from Rockwell Automation. Resource Description Industrial Automation Wiring and Grounding Guidelines, publication 1770-4.1 Provides general guidelines for installing a Rockwell Automation industrial system. Product Certifications website, http://www.ab.com Provides declarations of conformity, certificates, and other certification details. You can view or download publications at http://www.rockwellautomation.com/literature/. To order paper copies of technical documentation, contact your local Allen-Bradley distributor or Rockwell Automation sales representative. For Application on Bulletin 100/500/609/1200 Line Starters Heater Element Specifications Eutectic Alloy Overload Relay Heater Elements Type J — CLASS 10 Type P — CLASS 20 (Bul. 600 ONLY) Type W — CLASS 20 Type WL — CLASS 30 Note: Heater Element Type W/WL does not currently meet the material Type W Heater Elements restrictions related to EU ROHS Description The following is for motors rated for Continuous Duty: For motors with marked service factor of not less than 1.15, or Overload Relay Class Designation motors with a marked temperature rise not over +40 °C United States Industry Standards (NEMA ICS 2 Part 4) designate an (+104 °F), apply application rules 1 through 3. Apply application overload relay by a class number indicating the maximum time in rules 2 and 3 when the temperature difference does not exceed seconds at which it will trip when carrying a current equal to 600 +10 °C (+18 °F).
    [Show full text]
  • An Empirical Comparison of Pattern Recognition, Neural Nets, and Machine Learning Classification Methods
    An Empirical Comparison of Pattern Recognition, Neural Nets, and Machine Learning Classification Methods Sholom M. Weiss and Ioannis Kapouleas Department of Computer Science, Rutgers University, New Brunswick, NJ 08903 Abstract distinct production rule. Unlike decision trees, a disjunctive set of production rules need not be mutually exclusive. Classification methods from statistical pattern Among the principal techniques of induction of production recognition, neural nets, and machine learning were rules from empirical data are Michalski s AQ15 applied to four real-world data sets. Each of these data system [Michalski, Mozetic, Hong, and Lavrac, 1986] and sets has been previously analyzed and reported in the recent work by Quinlan in deriving production rules from a statistical, medical, or machine learning literature. The collection of decision trees [Quinlan, 1987b]. data sets are characterized by statisucal uncertainty; Neural net research activity has increased dramatically there is no completely accurate solution to these following many reports of successful classification using problems. Training and testing or resampling hidden units and the back propagation learning technique. techniques are used to estimate the true error rates of This is an area where researchers are still exploring learning the classification methods. Detailed attention is given methods, and the theory is evolving. to the analysis of performance of the neural nets using Researchers from all these fields have all explored similar back propagation. For these problems, which have problems using different classification models. relatively few hypotheses and features, the machine Occasionally, some classical discriminant methods arecited learning procedures for rule induction or tree induction 1 in comparison with results for a newer technique such as a clearly performed best.
    [Show full text]
  • Unsupervised Speech Representation Learning Using Wavenet Autoencoders Jan Chorowski, Ron J
    1 Unsupervised speech representation learning using WaveNet autoencoders Jan Chorowski, Ron J. Weiss, Samy Bengio, Aaron¨ van den Oord Abstract—We consider the task of unsupervised extraction speaker gender and identity, from phonetic content, properties of meaningful latent representations of speech by applying which are consistent with internal representations learned autoencoding neural networks to speech waveforms. The goal by speech recognizers [13], [14]. Such representations are is to learn a representation able to capture high level semantic content from the signal, e.g. phoneme identities, while being desired in several tasks, such as low resource automatic speech invariant to confounding low level details in the signal such as recognition (ASR), where only a small amount of labeled the underlying pitch contour or background noise. Since the training data is available. In such scenario, limited amounts learned representation is tuned to contain only phonetic content, of data may be sufficient to learn an acoustic model on the we resort to using a high capacity WaveNet decoder to infer representation discovered without supervision, but insufficient information discarded by the encoder from previous samples. Moreover, the behavior of autoencoder models depends on the to learn the acoustic model and a data representation in a fully kind of constraint that is applied to the latent representation. supervised manner [15], [16]. We compare three variants: a simple dimensionality reduction We focus on representations learned with autoencoders bottleneck, a Gaussian Variational Autoencoder (VAE), and a applied to raw waveforms and spectrogram features and discrete Vector Quantized VAE (VQ-VAE). We analyze the quality investigate the quality of learned representations on LibriSpeech of learned representations in terms of speaker independence, the ability to predict phonetic content, and the ability to accurately re- [17].
    [Show full text]
  • Training Generative Adversarial Networks in One Stage
    Training Generative Adversarial Networks in One Stage Chengchao Shen1, Youtan Yin1, Xinchao Wang2,5, Xubin Li3, Jie Song1,4,*, Mingli Song1 1Zhejiang University, 2National University of Singapore, 3Alibaba Group, 4Zhejiang Lab, 5Stevens Institute of Technology {chengchaoshen,youtanyin,sjie,brooksong}@zju.edu.cn, [email protected],[email protected] Abstract Two-Stage GANs (Vanilla GANs): Generative Adversarial Networks (GANs) have demon- Real strated unprecedented success in various image generation Fake tasks. The encouraging results, however, come at the price of a cumbersome training process, during which the gen- Stage1: Fix , Update 풢 풟 풢 풟 풢 풟 풢 풟 erator and discriminator풢 are alternately풟 updated in two 풢 풟 stages. In this paper, we investigate a general training Real scheme that enables training GANs efficiently in only one Fake stage. Based on the adversarial losses of the generator and discriminator, we categorize GANs into two classes, Sym- Stage 2: Fix , Update 풢 풟 metric GANs and풢 Asymmetric GANs,풟 and introduce a novel One-Stage GANs:풢 풟 풟 풢 풟 풢 풟 풢 gradient decomposition method to unify the two, allowing us to train both classes in one stage and hence alleviate Real the training effort. We also computationally analyze the ef- Fake ficiency of the proposed method, and empirically demon- strate that, the proposed method yields a solid 1.5× accel- Update and simultaneously eration across various datasets and network architectures. Figure 1: Comparison풢 of the conventional풟 Two-Stage GAN 풢 풟 Furthermore,풢 we show that the proposed풟 method is readily 풢 풟 풢 풟 풢 풟 training scheme (TSGANs) and the proposed One-Stage applicable to other adversarial-training scenarios, such as strategy (OSGANs).
    [Show full text]
  • Deep Neural Networks for Pattern Recognition
    Chapter DEEP NEURAL NETWORKS FOR PATTERN RECOGNITION Kyongsik Yun, PhD, Alexander Huyen and Thomas Lu, PhD Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, US ABSTRACT In the field of pattern recognition research, the method of using deep neural networks based on improved computing hardware recently attracted attention because of their superior accuracy compared to conventional methods. Deep neural networks simulate the human visual system and achieve human equivalent accuracy in image classification, object detection, and segmentation. This chapter introduces the basic structure of deep neural networks that simulate human neural networks. Then we identify the operational processes and applications of conditional generative adversarial networks, which are being actively researched based on the bottom-up and top-down mechanisms, the most important functions of the human visual perception process. Finally, Corresponding Author: [email protected] 2 Kyongsik Yun, Alexander Huyen and Thomas Lu recent developments in training strategies for effective learning of complex deep neural networks are addressed. 1. PATTERN RECOGNITION IN HUMAN VISION In 1959, Hubel and Wiesel inserted microelectrodes into the primary visual cortex of an anesthetized cat [1]. They project bright and dark patterns on the screen in front of the cat. They found that cells in the visual cortex were driven by a “feature detector” that won the Nobel Prize. For example, when we recognize a human face, each cell in the primary visual cortex (V1 and V2) handles the simplest features such as lines, curves, and points. The higher-level visual cortex through the ventral object recognition path (V4) handles target components such as eyes, eyebrows and noses.
    [Show full text]
  • Mixed Pattern Recognition Methodology on Wafer Maps with Pre-Trained Convolutional Neural Networks
    Mixed Pattern Recognition Methodology on Wafer Maps with Pre-trained Convolutional Neural Networks Yunseon Byun and Jun-Geol Baek School of Industrial Management Engineering, Korea University, Seoul, South Korea {yun-seon, jungeol}@korea.ac.kr Keywords: Classification, Convolutional Neural Networks, Deep Learning, Smart Manufacturing. Abstract: In the semiconductor industry, the defect patterns on wafer bin map are related to yield degradation. Most companies control the manufacturing processes which occur to any critical defects by identifying the maps so that it is important to classify the patterns accurately. The engineers inspect the maps directly. However, it is difficult to check many wafers one by one because of the increasing demand for semiconductors. Although many studies on automatic classification have been conducted, it is still hard to classify when two or more patterns are mixed on the same map. In this study, we propose an automatic classifier that identifies whether it is a single pattern or a mixed pattern and shows what types are mixed. Convolutional neural networks are used for the classification model, and convolutional autoencoder is used for initializing the convolutional neural networks. After trained with single-type defect map data, the model is tested on single-type or mixed- type patterns. At this time, it is determined whether it is a mixed-type pattern by calculating the probability that the model assigns to each class and the threshold. The proposed method is experimented using wafer bin map data with eight defect patterns. The results show that single defect pattern maps and mixed-type defect pattern maps are identified accurately without prior knowledge.
    [Show full text]
  • Introduction to Reinforcement Learning: Q Learning Lecture 17
    LLeeccttuurree 1177 Introduction to Reinforcement Learning: Q Learning Thursday 29 October 2002 William H. Hsu Department of Computing and Information Sciences, KSU http://www.kddresearch.org http://www.cis.ksu.edu/~bhsu Readings: Sections 13.3-13.4, Mitchell Sections 20.1-20.2, Russell and Norvig Kansas State University CIS 732: Machine Learning and Pattern Recognition Department of Computing and Information Sciences LLeeccttuurree OOuuttlliinnee • Readings: Chapter 13, Mitchell; Sections 20.1-20.2, Russell and Norvig – Today: Sections 13.1-13.4, Mitchell – Review: “Learning to Predict by the Method of Temporal Differences”, Sutton • Suggested Exercises: 13.2, Mitchell; 20.5, 20.6, Russell and Norvig • Control Learning – Control policies that choose optimal actions – MDP framework, continued – Issues • Delayed reward • Active learning opportunities • Partial observability • Reuse requirement • Q Learning – Dynamic programming algorithm – Deterministic and nondeterministic cases; convergence properties Kansas State University CIS 732: Machine Learning and Pattern Recognition Department of Computing and Information Sciences CCoonnttrrooll LLeeaarrnniinngg • Learning to Choose Actions – Performance element • Applying policy in uncertain environment (last time) • Control, optimization objectives: belong to intelligent agent – Applications: automation (including mobile robotics), information retrieval • Examples – Robot learning to dock on battery charger – Learning to choose actions to optimize factory output – Learning to play Backgammon
    [Show full text]
  • Image Classification by Reinforcement Learning with Two-State Q-Learning
    EasyChair Preprint № 4052 Image Classification by Reinforcement Learning with Two-State Q-Learning Abdul Mueed Hafiz and Ghulam Mohiuddin Bhat EasyChair preprints are intended for rapid dissemination of research results and are integrated with the rest of EasyChair. August 18, 2020 Image Classification by Reinforcement Learning with Two- State Q-Learning 1* 2 Abdul Mueed Hafiz , Ghulam Mohiuddin Bhat 1, 2 Department of Electronics and Communication Engineering Institute of Technology, University of Kashmir Srinagar, J&K, India, 190006 [email protected] Abstract. In this paper, a simple and efficient Hybrid Classifier is presented which is based on deep learning and reinforcement learning. Q-Learning has been used with two states and 'two or three' actions. Other techniques found in the literature use feature map extracted from Convolutional Neural Networks and use these in the Q-states along with past history. This leads to technical difficulties in these approaches because the number of states is high due to large dimensions of the feature map. Because our technique uses only two Q-states it is straightforward and consequently has much lesser number of optimization parameters, and thus also has a simple reward function. Also, the proposed technique uses novel actions for processing images as compared to other techniques found in literature. The performance of the proposed technique is compared with other recent algorithms like ResNet50, InceptionV3, etc. on popular databases including ImageNet, Cats and Dogs Dataset, and Caltech-101 Dataset. Our approach outperforms others techniques on all the datasets used. Keywords: Image Classification; ImageNet; Q-Learning; Reinforcement Learning; Resnet50; InceptionV3; Deep Learning; 1 Introduction Reinforcement Learning (RL) [1-4] has garnered much attention [4-13].
    [Show full text]