Deep Neural Network Tony Wong 2017-02-18 Outline ▸ Introduction to Neural Network ▸ Image Classifiers Using Tensorflow (Version 1.0) ▹ MNIST

Total Page:16

File Type:pdf, Size:1020Kb

Deep Neural Network Tony Wong 2017-02-18 Outline ▸ Introduction to Neural Network ▸ Image Classifiers Using Tensorflow (Version 1.0) ▹ MNIST Deep Neural Network Tony Wong 2017-02-18 Outline ▸ Introduction to Neural Network ▸ Image classifiers using Tensorflow (version 1.0) ▹ MNIST ▹ IOI Art Class ▹ GFRIEND dataset Deep Neural Network 2 Machine Learning ▸ Supervised Learning: ▹ Learn from examples to find patterns to explain the relationships between inputs and outputs ▸ http://playground.tensorflow.org ▸ Can you name some other applications that make use of machine learning? Deep Neural Network 3 Steps to Perform Machine Learning ▸ Collection: Define data structure and collect data ▸ Design: Select suitable algorithm ▸ Implementation: Build the computation graph by writing code ▸ Preprocessing: Split data into training and validation sets ▸ Training: Train the model using training examples ▸ Validation: How well does the model perform? ▸ Serving: Use the model to predict new data Deep Neural Network 4 Linear Regression ▸ When there are 0 hidden layers ▸ Find parameters θ h(x) = θ1x1+θ2x2 Deep Neural Network 5 Neural Network ▸ A neural network can help us find non-linear patterns in the data ▸ Example: y = (x < 0) XOR (y < 0) Deep Neural Network 6 c = sum{xi * wi} + b bias weight y = f(c) Neuron x1 1 activation function weight2 c y x 2 ▸ Using an activation function ReLU: y = max(0, c) introduces non-linearity, which Rectified Linear Unit can help the neural network to learn sigmoid: y = 1/(1-e-x) tanh: y = 2/(1-e-2x)-1 Deep Neural Network 7 Loss ▸ Error rate ▸ Machine learning is about minimizing the loss by adjusting parameters (weights) loss Gradient Descent Deep Neural Network 8 Loss != Accuracy ▸ Loss decreases when Which model is better? ▹ Accuracy increases Model 1 Model 2 T 20% 90% ▹ Correct predictions are Q1 made with higher confidence F 80% 10% T 52% 49% ▹ Incorrect predictions are Q2 F 48% 51% made with lower confidence T 47% 52% Q3 F 53% 48% Accuracy 66.7% 33.3% Deep Neural Network 9 Random initialization ▸ The weights are random initialized so that each neuron can detect different patterns ▸ Different initializations can arrive at different local minima Deep Neural Network 10 Training and Validation (Testing) Loss ▸ We split the examples into 2 sets: training and validation ▸ For example, if we have 1000 examples ▹ We can make 800 training examples for the network to learn ▹ 200 validation examples to evaluate the network's ability to explain unseen examples Deep Neural Network 11 All input features, 3 hidden layers: 8-6-4 Underfitting and Overfitting Overfit: gap between test and training loss x and y features, x and y features, 1 hidden layer: 2 1 hidden layer: 4 Underfit: high training Just right: low training Deep Neural Network 12 and validation losses and validation losses Underfitting and Overfitting ▸ To fix underfitting ▹ More input features ▹ More hidden layers ▸ To fix overfitting ▹ More training examples ▹ Use simpler model ▹ Add regularization Deep Neural Network 13 MNIST Database ▸ Mixed National Institute of Standards and Technology ▸ Multiclass-classification of hand-written digits ▸ Each example in MNIST consists of the 28x28 image and the corresponding label (0-9) ▸ If we express the label using one-hot encoding, it would be a 1D tensor of shape [10] ▹ [0,0,0,0,0,0,1,0,0,0] for digit 6 CC-BY 3.0: https://www.tensorflow.org/versions/master/tutorials/mnist/beginners/ Deep Neural Network 14 MNIST Database ▸ The database contains 50000 training examples (for learning) and 10000 validation examples (for evaluation) ▸ Can you guess what's the validation accuracy achieved by the best model? (random guessing = 10%) ▸ What about by the average human? <80% 80-90% 90-95% 95%-97% 97-98% 98-99% 99-99.5% >99.5% Deep Neural Network 15 Tensorflow ▸ The flow of data is a directed acyclic graph ▸ Tensor (multidimensional matrices) flows through edges ▸ Nodes are operators (e.g. load data, addition, matrix multiplication, max) ▸ Write Python/C++ code to describe the graph ▸ Tensorflow manages the execution (e.g. distribute computation to different CPU / GPU) ▸ pip install --upgrade tensorflow Deep Neural Network 16 Tensor ▸ Example of 1D example: ▹ height weight age loc_nt loc_kln loc_hk ▹ 158 50 12 0 1 0 (one-hot encoding for location) ▸ 2D examples: grayscale images ▹ 28(h) x 28(w) ▸ 3D examples: RGB images ▹ 32(h) x 32(w) x 3(channels) CC-BY 3.0: https://www.tensorflow.org/versions/master/tutorials/mnist/beginners/ Deep Neural Network 17 Tensor ▸ When we stack 55000 28x28 grayscale images together ▹ We have a tensor of shape [55000, 28, 28,1] ▹ We can also reshape the tensor to [55000, 784] CC-BY 3.0: https://www.tensorflow.org/versions/master/tutorials/mnist/beginners/ Deep Neural Network 18 Windows Remote Desktop Connection ▸ A virtual machine has been set up with Tensorflow installed ▸ Windows -> type "Remote" ▸ Hostname: hkoi11530.cloudapp.net Deep Neural Network 19 File Templates and Datasets ▸ Some templates and datasets have been prepared for you ▸ Go to the Public and copy tensorflow, tfpreprocess and mystery.7z to your user's Documents Deep Neural Network 20 Notes ▸ In the next steps we are only defining the graph ▹ Nothing is computed yet ▸ In Python, indentation defines the scope ▹ Each line in a scope must be indented in the same way ▹ Mixing tabs with spaces results in error ▹ Since Tensorflow is developed by Google, which uses 4 spaces, you should use 4 spaces as well Deep Neural Network 21 b1 a1 w11 Creating a Neural Network w12 c1 f1 a2 ▸ Let's build a network with 1 hidden layer ci = sum{aj* wij} + bi fi = max(0, ci) ▹ Activation function: ReLU Input Layer Bias Hidden Layer Bias Output Layer ... 10 nodes ... ... (1 for each digit) 28x28 nodes ?? nodes ... (1 for each pixel) (1 for each kernel) ... ... Deep Neural Network 22 Creating a Neural Network ▸ Create a file simple.py Returns Input Tensor unknown number of images images y inference [-1, 28, 28, 1] [-1, 10] num_classes image_size channels Deep Neural Network 23 Bias Layer Architecture of a Layer 4 spaces ... ... input_size output_size nodes nodes ... ... biases [-1, input_size] [output_size] input_tensor local activations matmul + ReLU returns [-1, output_size] weights [input_size, output_size] [-1, output_size] Deep Neural Network 24 Stacking the Layers 4 spaces image_size = 28 channels = 1 HIDDEN_NODES = 48 num_classes = 10 reshaped hidden output [-1, 28, 28, 1] [-1, 784] [-1, 48] [-1, 10] images reshape layer(784, 48) layer(48, 10) returns Deep Neural Network 25 Interpreting the Output ▸ We predict the class having the highest value in the output inference main.py (written for you already) Deep Neural Network 26 Probabilities ▸ Probabilities should add up to 1 ▸ We apply softmax to the output Unlikely to be 8 y[i] y[i] ▸ P(y_ = i) = e / sum{e } The digit has 37% chance to be "2" Deep Neural Network 27 Loss function ▸ Compute the loss for the optimizer to train our model Shape: [-1] Rows containing cross entropy loss of each example Shape: [1] Average of the cross entropy losses Deep Neural Network 28 Cross Entropy ▸ For classification, one popular loss function is the Cross Entropy loss ▸ It takes account into the probabilities instead of accuracy Cross Entropy loss = −log (0.375) ≈ 0.98 (lower is better) Deep Neural Network 29 Optimizer ▸ The optimizer traverses the graph in reverse topological order to compute gradients at each node 훿푧 훿푧 훿푦 ▸ = 훿푥 훿푦 훿푥 ▸ Then the optimizer adjust the weights variables to lower the loss ▸ Different of optimizers adjust the weights differently Deep Neural Network 30 Notes ▸ These parts have been written for you ▹ Data loading (it has to be efficient because the data loading part is usually the bottleneck for small networks) ▹ Progress logging (accuracy and loss) ▹ Saving the model so that it can be restored for more training or serving Deep Neural Network 31 Training ▸ Open command prompt in the working directory ▸ python main.py --logdir log/mnist1 What's the accuracy at Step 1000? Training Validation Deep Neural Network 32 Visualizing the Training Progress ▸ Tensorflow comes with Tensorboard – a visualization tool ▸ Open another command prompt in the working directory ▸ tensorboard --logdir log --reload_interval 15 --port PORT ▸ Open the browser and enter URL: http://localhost:PORT Stop the training (CTRL+C) when the validation loss decreases very slowly Deep Neural Network 33 Confusion Matrix ▸ Confusion Matrix helps us to understand how well the model is performing, and which classes are particular difficult Deep Neural Network 34 Stopping, Resuming and Resetting Training ▸ Training will resume provided that you specified the same logdir ▹ The graph must be the same ▸ If you would like to reset and restart training using the same logdir, you must: ▹ Stop tensorboard (CTRL+C) otherwise you cannot delete the logdir ▹ Delete the logdir (e.g. mnist1 in log) ▹ Start training ▹ Start tensorboard again Deep Neural Network 35 Visualizing the Hidden Layer ▸ By visualizing the nodes in the Hidden Layer, we can have a sense of what patterns the neural network is trying to recognize ▸ Beware of indentation! Deep Neural Network 36 Visualizing the Hidden Layer ▸ python main.py --logdir log/mnist2 ▸ In Tensorboard, go to the Images tab ▸ Here you can see the weights of the 48 nodes in the hidden layer Deep Neural Network 37 Adding Regularization ▸ We add a L2 loss ▹ The squared weights will be added to the loss function ▹ The regularization controls the strength of regularization (0 = no regularization) ▸ python main.py --logdir log/mnist3 --regularization What's the accuracy at Step 1000? 0.005 Training Validation Deep Neural Network 38 Effects
Recommended publications
  • 2017 Mini Competition 0 M1701-5: Tony Wong M1706: Alex Tung (Story by Tony Wong) Yuju Sinb /News/End?Id=7941357
    2017 Mini Competition 0 M1701-5: Tony Wong M1706: Alex Tung (story by Tony Wong) http://m.star.naver.com/GFRIEND Yuju SinB /news/end?id=7941357 Eunha Yerin Sowon “ Umji M1701 - Hearts https://youtu.be/bwTVerz9X3c 2017 Mini Competition 0 3 M1701 - Hearts ▸ Each small heart and large heart requires exactly 1 Buddy ▸ Each extra large heart requires 2 extra Buddies ▸ These 2퐷 Buddies can make 2퐷 mini hearts ▹ 퐴 ← max⁡(0⁡, 퐴 − 2퐷) ▸ Finally, 퐴/2 additional Buddies are required to make the remaining mini hearts ▸ long long is required 2017 Mini Competition 0 4 M1702 – Archery http://ent.mbc.co.kr/content/4918 2017 Mini Competition 0 5 M1702 - Archery ▸ Compute the Euclidean distance of the center of the arrow to the origin: 푑 = 푥2 + 푦2 ▸ The score of the arrow is the max 푝 ∈ 1, 2, 3, … , 10 such that 푑 − 푟 ≤ 40(11 − 푝) ▸ To avoid precision error, we square both sides of the inequality 푥2 + 푦2 ≤ [40 11 − 푝 + 푟]2 ▸ To further avoid precision error, we store 푥, 푦 and 푟 using integers by multiplying the numbers by 10 (i.e. 푟 = 12⁡for 1.2mm) 푥2 + 푦2 ≤ [400 11 − 푝 + 푟]2 2017 Mini Competition 0 6 M1703 – Photo Collage Jeremy's favourite photo http://childyenni.tistory.com/11 2017 Mini Competition 0 7 M1703 – Photo Collage ▸ How to represent the photos? ▹ Sowon, Yerin, Eunha, Yuju, SinB, Umji ▹ Method 1: Index the members using 0, 1, 2, 3, 4, 5. Create a 2D boolean array A of size N x 6. Set A[i][j] to true if member j is present in photo i ▹ Method 2 (better): Index the members using 1, 2, 4, 8, 16, 32.
    [Show full text]
  • Business Proposal
    BUSINESS PARTNERSHIP PROPOSAL “FIRST TIME IN INDIA” New Delhi, INDIA 21st - 23rd AUGUST, 2020 ORGANISED BY NEXODE MEDIA PVT. LTD. ABOUT GLOBAL YOUTH LEADERS MODEL UNITED NATIONS Global Youth Leaders MUN (GYLMUN) will be held on August 21-23, 2020 in New Delhi, India. GYLMUN is aimed to provide a platform for youth to learn about diplomacy, critical thinking, public speaking and the United Nations Conference. The ultimate goal of GYLMUN is to encourage the youth to be aware of the international issues, understand and try to form a possible solution to solve particular issues related to the 17 Global Goals. The youth will feel the ambience of being representatives of their assigned countries and experience how the United Nations Conference executes their ideas and plans. It is the best platform where they can improve their soft skills and knowledge and expand their network. ABOUT GLOBAL YOUTH LEADERS MODEL UNITED NATIONS As an important part of the world, youth have responsibilities and rights to contribute to the realization of Global Goals as a key to transform our world into a better place to live in. As a youth capacity development platform, Nexode Media has consistently put effort to arrange some programs that are relevant to youth in today's need and for the better future of the world. The organization is eager to create a platform, form an alliance between young leaders, which will accommodate ideas from youth spread all over the countries through the programs. Youth leaders will get more perspectives from the world, thus, they will enhance more understanding about related issues.
    [Show full text]
  • UC Riverside Electronic Theses and Dissertations
    UC Riverside UC Riverside Electronic Theses and Dissertations Title K- Popping: Korean Women, K-Pop, and Fandom Permalink https://escholarship.org/uc/item/5pj4n52q Author Kim, Jungwon Publication Date 2017 Peer reviewed|Thesis/dissertation eScholarship.org Powered by the California Digital Library University of California UNIVERSITY OF CALIFORNIA RIVERSIDE K- Popping: Korean Women, K-Pop, and Fandom A Dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy in Music by Jungwon Kim December 2017 Dissertation Committee: Dr. Deborah Wong, Chairperson Dr. Kelly Y. Jeong Dr. René T.A. Lysloff Dr. Jonathan Ritter Copyright by Jungwon Kim 2017 The Dissertation of Jungwon Kim is approved: Committee Chairperson University of California, Riverside Acknowledgements Without wonderful people who supported me throughout the course of my research, I would have been unable to finish this dissertation. I am deeply grateful to each of them. First, I want to express my most heartfelt gratitude to my advisor, Deborah Wong, who has been an amazing scholarly mentor as well as a model for living a humane life. Thanks to her encouragement in 2012, after I encountered her and gave her my portfolio at the SEM in New Orleans, I decided to pursue my doctorate at UCR in 2013. Thank you for continuously encouraging me to carry through my research project and earnestly giving me your critical advice and feedback on this dissertation. I would like to extend my warmest thanks to my dissertation committee members, Kelly Jeong, René Lysloff, and Jonathan Ritter. Through taking seminars and individual studies with these great faculty members at UCR, I gained my expertise in Korean studies, popular music studies, and ethnomusicology.
    [Show full text]
  • Travail Personnel Korean Entertainment
    Travail Personnel Korean Entertainment Nom: Alves Chambel Prénom: Carolina Classe: 6G3 Tutrice: Amélie Mossiat 2019/2020 Introduction Hello, my name is Carolina and I am 14 years old. My 2019/2020 TRAPE is about Korean Entertainment! I have been into K-Pop for almost 3 years and I have always been researching lots of things about Korean celebrities, their activities and their TV shows. Since I was a child, I have always been very curious about the Asian culture, even though I thought they were gangster. As I grew up, I have become more and more connected to their culture. Then, at the end of 2016, I listened to my first K-Pop song ever. I was 10 years old and I was in primary school when I got to know EXO, the first group I listened to. They had a big impact in my life because in that moment I was struggling mentally and physically... The only thing that made me happy, was listening to their songs! They would make feel better in every way possible, once I listened to some music, I’d completely forget about my worries. It was always the best part of my days. That is why I have grown attached to K-Pop songs, dances and singers. Table of contents WHAT WERE THE MOST FAMOUS SERIES (KOREAN DRAMAS)? (2019) IS IT POPULAR IN OTHER COUNTRIES? WHO IS MY FAVOURITE ACTOR AND ACTRESS? ARE IDOLS ALSO ACTORS? WHAT GENRES OF DRAMAS ARE THERE? WHAT GENRE OF MUSIC DO THEY DO? WHAT ARE THE MOST FAMOUS GROUPS IN KOREA AND INTERNATIONALLY? WHAT ARE THE MOST FAMOUS SONGS? DO K-DRAMAS ALSO HAVE SONGS? What were the most famous series (Korean Dramas)? (2019) Since we are at the beginning of 2020, it would be better to make a list of the 10 most popular K- Dramas in 2019.
    [Show full text]
  • Anne Frank's Di
    Pakistan School Attack On the 17th December 2014, a greatly upsetting attack took place in Pakistan. Taliban terrorists shot 132 children [aging from twelve to six- teen], ten school staff members, including the principal of the school, and three soldiers in the Army Public School and Degree College. A hundred people were also injured, many having gunshot wounds. It was found out that the terrorists burst into the auditorium where a large number of students were taking an exam and gunned down many of them within minutes. A 14-year-old survivor, Ahmed Faraz, recalled that one of the terrorists ordered his men to kill all the children hiding under the benches. They carried on shooting incessantly. Parents send their children to school expecting them to be safe. How- ever in some countries, school is not as safe as schools in Singapore. So, we should stay alert and be thankful to have this privilege for the safe en- vironment where we study in. It was also said that the students thought that the attack was a drill. As a result, most of them did not take it very seriously. Having said that, many of us tend to take drills in school lightly by joking around and thinking that it is a waste of time. These attacks can be avoided if all of us cooperate and stay united. The root cause of such at- tacks is based on a religion not being accepted by others around them. In this article, we observe that the Islamic culture was not accepted and thus these attacks occurred.
    [Show full text]
  • Thần Tượng Biết Yêu (Tfboys Version)
    Thần Tượng Biết Yêu (Tfboys Version) Contents Thần Tượng Biết Yêu (Tfboys Version) 2 1. Chương 1: Quá Khứ Đau Thương ................................... 3 2. Chương 2: Cuộc Sống Mới ....................................... 7 3. Chương 3: Buổi Tiệc .......................................... 7 4. Chương 4: Buổi Tiệc (phần 2) ..................................... 9 5. Chương 5: Buổi Tiệc (phần 3) ..................................... 13 6. Chương 6: Buổi Tiệc (phần Cuối) ................................... 15 7. Chương 7: Lại Chạm Mặt ....................................... 18 8. Chương 8: Mẹ Tái Hôn Ư (phần 1) .................................. 20 9. Chương 9: Mẹ Tái Hôn Ư (phần 2) .................................. 22 10. Chương 10: Lên Truyền Hình Cùng Nhau (phần 1) ......................... 24 11. Chương 11: Lên Truyền Hình Cùng Nhau(phần 2) ......................... 27 12. Chương 12: Lên Truyền Hình Cùng Nhau (phần 3) ......................... 29 13. Chương 13: Đi Học (phần 1) ..................................... 31 14. Chương 14: Đi Học (phần 2) ..................................... 32 15. Chương 15: Căn-tin(phần 1) ..................................... 34 16. Chương 16: Căn-tin(phần 2) ..................................... 34 17. Chương 17: Gặp Lại Nhau,tình Yêu Đầu Đời Của Băng(phần 1) . 36 18. Chương 18: Gặp Lại Nhau, Tình Yêu Đầu Đời Của Băng (phần 2) . 37 19. Chương 19: Chị Hai Của Trường Đã Về ............................... 38 20. Chương 20: Angel Đã Đến ...................................... 40 21. Chương 21 ..............................................
    [Show full text]
  • GPMS Gazette March, 2019
    The Official Newsletter of Gregory-Portland Middle School: Home of the World’s Greatest Students! Academic UIL winners: Brodie Mitchel, Diego Aguillon, Adrian Galvan, Natalie DeLeon, Emma Denton & Erin Ebers. More details inside! Photo by Brooke Moreno Visit www.g-pisd.org/gpms/community to view, GPMS Gazette download, and/or print your own copies in… March, 2019 Brought to you by students in the GPMS Press Corps! Jonathon Martinez Executive Editor March “color” by Jessie Riojas GPMS NEWS News Editor: Faith Pitts (with help from Toonie, the Press Corps Broonie) Math Counts By Melany Castillo Photos provided by Collen Johnson 3rd-place winners (left-to-right, above) Lleyton Davidson, Elisabeth Miller, Eli Gerick and Elena Miller represented GPMS at the February 2, 2019 Math Counts meet. The team was selected by Mrs. Biediger, and some volunteered. The team can only have 4 students. Mrs. Biediger selected them from her UIL team. There were additional students who went to the November practice round. Eight students participated in the practice meet last fall. (Continued on next page) (Math Counts, continued from previous page) The Nueces Chapter of the Texas Society of Professional Engineers and some local companies sponsor the event each year to challenge 6th-8th grade students with engineering-type problem solving. The competition consists of several different tests and a team test, like the UIL Math competitions. Currently they provide a stipend of up to $500 each competition year to up to two teachers per school for attendance at a free workshop in the fall with continuing education credits ($100), the practice ($200) and the competition ($200).
    [Show full text]
  • Korean Women, K-Pop, and Fandom a Dissertation Submitted in Partial Satisfaction
    UNIVERSITY OF CALIFORNIA RIVERSIDE K- Popping: Korean Women, K-Pop, and Fandom A Dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy in Music by Jungwon Kim December 2017 Dissertation Committee: Dr. Deborah Wong, Chairperson Dr. Kelly Y. Jeong Dr. René T.A. Lysloff Dr. Jonathan Ritter Copyright by Jungwon Kim 2017 The Dissertation of Jungwon Kim is approved: Committee Chairperson University of California, Riverside Acknowledgements Without wonderful people who supported me throughout the course of my research, I would have been unable to finish this dissertation. I am deeply grateful to each of them. First, I want to express my most heartfelt gratitude to my advisor, Deborah Wong, who has been an amazing scholarly mentor as well as a model for living a humane life. Thanks to her encouragement in 2012, after I encountered her and gave her my portfolio at the SEM in New Orleans, I decided to pursue my doctorate at UCR in 2013. Thank you for continuously encouraging me to carry through my research project and earnestly giving me your critical advice and feedback on this dissertation. I would like to extend my warmest thanks to my dissertation committee members, Kelly Jeong, René Lysloff, and Jonathan Ritter. Through taking seminars and individual studies with these great faculty members at UCR, I gained my expertise in Korean studies, popular music studies, and ethnomusicology. Thank you for your essential and insightful suggestions on my work. My special acknowledgement goes to the Korean female K-pop fans who were willing to participate in my research.
    [Show full text]
  • Semiotika Dalam Novel I Am Sarahza Karya Hanum Salsabiela Rais Dan Rangga Almahendra” Dengan Baik
    SEMIOTIKA DALAM NOVEL I AM SARAHZA KARYA HANUM SALSABIELA RAIS DAN RANGGA ALMAHENDRA SKRIPSI Disusun untuk Memperoleh Gelar Sarjana Pendidikan pada Universitas Islam Sultan Agung Oleh Fitria Tamaroh 34101400143 PROGRAM STUDI PENDIDIKAN BAHASA DAN SASTRA INDONESIA GAKULTAS KEHURUAN DAN ILMU PENDIDIKAN UNIVERSITAS ISLAM SULTAN AGUNG SEMARANG 2019 MOTTO DAN PERSEMBAHAN 1. Allah tidak akan membebani seseorang melainkan sesuai dengan kadar kesanggupannya (Al Baqarah : 286) 2. Hadapi rasa takut itu tanpa memikirkan apa yang akan menimpamu selanjutnya. Selama apa yang kau lakukan itu benar, maka rasa takut itu hanyalah sebuah kerikil yang mengganggu. Terasa sedikit perih saat dilempar padamu, namun bahkan rasa pedih itu sama sekali tak terasa setelah satu detik berlalu. 3. Nothing impossible as long as believe. Stay strong and do everything you want. Cause Allah will give all of you need. PERSEMBAHAN Skripsi ini penulis persembahkan kepada Bapak Sumarlan dan Ibu Suntamah, orang tua saya yang senantiasa mendampingi saya dalam menyelesaikan karya ilmiah ini, menyelipkan nama saya dalam setiap sujud beliau, memberikan dukungan serta motivasi untuk saya untuk berjuang menyelesaikan karya ilmiah ini tanpa melupakan ibadah saya juga. Untuk adik saya tercinta, Ma’ruf Mahmuda yang menjadi sumber kekuatan saya ketika saya merasa lelah. PRAKATA Tiada kata atau kalimat indah yang dapat saya sampaikan terkecuali lafal Alhamdulillahirobbilalamin sebagai wujud syukur kepada Allah Swt, yang telah melimpahkan rahmat dan hidayahnya kepada peneliti sehingga dapat menyelesaikan skripsi “Semiotika dalam Novel I Am Sarahza Karya Hanum Salsabiela Rais dan Rangga Almahendra” dengan baik. Skripsi disusun sebagai bentuk memenuhi persyaratan dalam meraih gelar Sarjana pada Program Bidang Studi Pendidikan Bahasa dan Sastra Indonesia, Universitas Islam Sultan Agung.
    [Show full text]