Crash Course Artificial Intelligence Episodes #1-5 PDF

Crash Course Artificial Intelligence Episodes #1-5 What is Artificial Intelligence? Supervised Learning Neural Networks and Deep Learning Training Neural Networks How to Make an A.I. read your Ha...

Crash Course Artificial Intelligence Episodes #1-5 What is Artificial Intelligence? Supervised Learning Neural Networks and Deep Learning Training Neural Networks How to Make an A.I. read your Handwriting (Lab) Includes Complete Answer Key  Check out my other Crash Course Guides and Products in my Store https://www.teacherspayteachers.com/Store/Karen-Maxson Crash Course Guides: Novel Studies: History of Science The Voyage of the Frog U.S. History By Gary Paulsen Navigating Digital Information Statistics Night journeys Theater & Drama By Avi Biology World History Boy at War Business By Harry Mazer European History Engineering I Survived True Stories Literature By Lauren Tarshis Government and Politics A.I. (Artificial Intelligence)  Thanks for purchasing my product and supporting my business! Questions? Earn TPT Credits! Please email me! Remember to leave feedback [email protected] to earn TPT credit! Look for more resources and episode guides in my store: https://www.teacherspayteachers.com/Store/Karen-Maxson Terms of Use: By purchasing this resource, you are agreeing that the contents are the property of Karen Maxson and licensed to you only for classroom/personal use as a single user. I retain the copyright, and reserve all rights to this product. You may not: claim this work as your own, alter the files in any way, remove the copyright, or sell or post for free any part of this work. Questions about usage? Please contact me: [email protected]. Clip Art Thanks to:  Episode Links What is Artificial Intelligence? https://www.youtube.com/watch?v=a0_lo_GDcFw Supervised Learning https://www.youtube.com/watch?v=4qVRBYAdLAo Neural Networks and Deep Learning https://www.youtube.com/watch?v=oV3ZY6tJiA0 Training Neural Networks https://www.youtube.com/watch?v=lgKrup5oi_A How to Make an A.I. read your Handwriting (Lab) https://www.youtube.com/watch?v=6nGCGYWMObE&t=181s  Table of Contents Episode Guides with time-stamps…………………………………………………………………………………………………………….………6 Ep. 1: “What is Artificial Intelligence?”……………………………………………………….………………...………….…7 Ep. 2: “Supervised Learning”...............................................................................................................................9 Ep. 3: “Neural Networks and Deep Learning” ….……..……………………..……………..…………….1 1 Ep. 4: “Training Neural Networks”............................................................................................................12 Ep. 5: “How to Make an A.I. read your Handwriting (Lab)” ……………...……….13 Episode Guides without timestamps……………………………………………………………………………………...…………..………..15 Ep. 1: “What is Artificial Intelligence?”…………………………..…………………………………………………….….…16 Ep. 2: “Supervised Learning”.............................................................................................................................18 Ep. 3: “Neural Networks and Deep Learning” ….……..………………….…………...……………...20 Ep. 4: “Training Neural Networks”............................................................................................................21 Ep. 5: “How to Make an A.I. read your Handwriting (Lab)” ……………..……..22 Answer Keys………………………………………………………………………………………………………………………………………………………....….……...…24 Ep. 1: “What is Artificial Intelligence?”………………………………...……………………………………………...…25 Ep. 2: “Supervised Learning”..........................................................................................................................26 Ep. 3: “Neural Networks and Deep Learning” ….……..……………………………………...……….27 Ep. 4: “Training Neural Networks”........................................................................................................28 Ep. 5: “How to Make an A.I. read your Handwriting (Lab)” ……………..……..29  Episode Guides With Timestamps https://www.teacherspayteachers.com/Store/Karen-Maxson  Name: ________________________ Crash Course Artificial Intelligence Episode #1 : What is Artificial Intelligence? 1. Which of the following use A.I.? (0:10) (Circle all that apply) A. Banks C. YouTube B. Doctors D. Apps for cell phones 2. How do you feel about A.I.? (0:30) ____________________________________________ ____________________________________________________________________ ____________________________________________________________________ 3. “Writers like to imagine a more _________________________ A.I.” (1:30) 4. This is a realistic way to think of A.I. (1:35) TRUE / FALSE 5. “A machine is said to have artificial intelligence if it can _____________________________, potentially __________________ from the data, and use that knowledge to _____________ and achieve specific goals.” (1:50) 6. Today’s A.I. still can’t do most of the things that humans do. (3:38) TRUE / FALSE 7. A.I. is used to screen ____________________ applications. (4:21) Do you think this is fair? Explain. ____________________________________________________________________ ____________________________________________________________________ ____________________________________________________________________ 8. When was the term “artificial intelligence” coined? (4:55) ____________________________ 7 © 2019, K. Maxson  9. What important conference is responsible for a lot of advancements in A.I.? (5:00) ____________________________________________________________________ 10. A.I. has developed FASTER / SLOWER than these early thinkers predicted. (5:47) 11. When did the “A.I. Winter” end? (7:15) ________________________________________ 12. What “two big developments” helped to spur the A.I. revolution forward? 1. Increase in ____________________________ (7:43) 2. __________________ and ________________________(9:50) 13. What increases the speed of a computer? (8:35) A. More hard drives C. Faster internet speed B. More circuit boards D. More transistors 14. When was the first iPhone released? (8:48) A. 2001 C. 2007 B. 2005 D. 2011 Direct Episode Link: https://www.youtube.com/watch?v=a0_lo_GDcFw 8 © 2019, K. Maxson  Name: ________________________ Crash Course Artificial Intelligence Episode #2: Supervised Learning 1. What process helps computers learn? (0:30) _____________________________________ 2. What are the “three main types of learning?” (1:03) ____________________________________________________________________ 3. “Supervised learning is the process of learning with _______________________________.” (1:35) 4. What two things does an A.I. need to learn? (2:05) A. Humans to supervise C. Answer key B. Computing power D. Data 5. Scientists were inspired by the human brain when they made some kinds of A.I. (2:45) TRUE / FALSE 6. What are the three basic parts of a neuron? (2:55) ____________________________________________________________________ 7. What machine did Rosenblatt build? What was he trying to teach it to do? (3:45) ____________________________________________________________________ 8. What special weight represents the threshold? (5:35) _______________________________ 9. Generalized A.I. was successfully created in the early 2010’s. (6:40) TRUE / FALSE 10. Name 2 different activation functions. (7:25) _____________________________________ 11. At the beginning of supervised learning, the A.I. makes random guesses. (8:58) TRUE / FALSE 9 © 2019, K. Maxson  12. When John Green Bot is correct, the update rule will add “__________ to the weight.” (9:25) 13. John Green Bot learns from SUCCESS / FAILURE. (9:45) 14. You can put the results of a supervised learning test into “a table, called a _________________ ______________________.” (11:50) 15. “____________________ tells us how much you should _____________ your program when it says it’s found something.” (12:25) 16. “____________________ tells you how much your program can find of the thing you’re looking for.” (12:48) 17. What is the “key to most A.I. challenges? (13:53) __________________________________ Direct Episode Link: https://www.youtube.com/watch?v=4qVRBYAdLAo 10 © 2019, K. Maxson  Name: ________________________ Crash Course Artificial Intelligence Episode #3: Neural Networks and Deep Learning 1. How many neurons does our brain have? (0:15) A. 100 thousand C. 100 billion B. 100 million D. 100 trillion 2. “Not that long ago, a big challenge in AI was ____________________________ recognition.” (0:52) 3. Who helped to develop AI that could recognize images? (1:24) _________________________ 4. What was their international dataset called? How many images does it have? (1:43) ____________________________________________________________________ 5. What made AlexNet different? (2:40) _________________________________________ 6. “All neural networks are made up of an ______________ layer, an ________________ layer, and any number of _______________ layers in between.” (3:24) 7. Each pixel can be represented with 3 numbers that correspond to the amount of ____________, _______________, and ______________ in the pixel. (4:22) 8. What layer is “where the final hidden layer outputs are mathematically combined to answer the problem?” (5:12) ____________________________________________________________________ 9. What is “the key to neural networks and…all of A.I.?” (5:47) ___________________________ 10. Every hidden neuron looks for all of the components needed to determine if the picture is what they are looking for. (i.e. a dog, or cat) (7:30) TRUE / FALSE 11. How many neurons are in the output layer? (8:34) _________________ 12. Deeper neural networks “are networks with more _________________________ to do deep _________________.” (9:42) 13. What is one way that neural networks are used in your life already? (10:45) ________________ ____________________________________________________________________ Direct Episode Link: https://www.youtube.com/watch?v=oV3ZY6tJiA0 11 © 2019, K. Maxson  Name: ________________________ Crash Course Artificial Intelligence Episode #4: Training Neural Networks 1. Neural networks are now big enough that some can do tasks better than humans. (0:12) TRUE / FALSE 2. What algorithm do neural networks use to handle mistakes? (0:40) _____________________ 3. What are the “two main parts” of neural networks? (0:50) ____________________________ 4. “The task of finding the best weights for a neural network architecture is called ______.” (1:12) A. Backpropagation C. Nodular B. Optimization D. Synapse 5. What is “the goal of linear regression?” (2:12) ____________________________________ ____________________________________________________________________ 6. “By connecting together many simple _______________ with __________________, a neural network can learn to solve complicated _____________________.” (3:10) 7. What do neurons multiply the given features by? (4:30) _____________________________ 8. What represents the error when “the predicted answer and the correct answer is more than just one number?” (5:10) ____________________________________________________________________ 9. What algorithm did scientists create “to help neural networks learn?” (5:42) ____________________________________________________________________ 10. A _____________________________ is “where the weights make the error relatively small, but not the smallest it could be.” (8:18) 11. Ideally, you should explore different solutions at the same time, on the same neural network. However, that just isn’t feasible yet. (9:10) TRUE / FALSE 12. “Over time, backpropagation will adjust the neuron’s _________________, so that the neural networks output matches the ________________________...That’s called _____________ _________________________________________.” (10:38) Direct Episode Link: 12 https://www.youtube.com/watch?v=lgKrup5oi_A © 2019, K. Maxson  Name: ________________________ Crash Course Artificial Intelligence Episode #5: how to Make an AI read your handwriting (LAB) 1. In this episode “we’ll tackle a ________________ together and program a ___________________________ to recognize __________________________.” (1:15) 2. What language will be used to write the code in this episode? (1:30) A. JavaScript C. C# B. Python D. PHP 3. Where can you find the code that the episode is working on today? (1:35) ____________________________________________________________________ 4. “Our goal is to program a neural network to recognize _____________________ letters and convert them to ______________________________.” (2:10) 5. “Usually with a project like this, we’d have to write code to figure out where one letter ends and another begins...That’s called the _______________________________________.” (2:25) 6. Step 1: “Find or create a ______________________________ to train our neural network.” (3:17) 7. Step 2: “Create a ________________________________.” (3:35) 8. Step 3: “Train, _________________, and tweak our code until we feel that it’s ___________________ enough.” (3:48) Step 4: Scan John-Green-Bot’s handwritten pages and use our newly trained neural network to convert them into typed text!” 9. What dataset (abbreviated) will be used for this lab? (4:15) ___________________________ 10. Step 1.2 includes training the network with 100,000 images. (5:00) TRUE / FALSE 13 © 2019, K. Maxson  11. “___________________________________ is a very important step to training a good neural network.” (5:57) 12. What type of neural network will they be using in their episode? (6:50) ____________________________________________________________________ 13. “The size of our output layer depends on the number of __________________________ that we want our neural network to guess.” (7:43) 14. How many output neurons will this network have? (7:50) A. 26 C. 130 B. 52 D. 1,300 15. What number does the program print for every epoch? (9:50) ____________________________________________________________________ You want to see this number go UP / DOWN over time. 16. What is the tradeoff if you add more layers and neurons to your network to make it more accurate? (11:20) ____________________________________________________________________ 17. Where are the images for this step being stored? (12:47) _____________________________ 18. Does the program process the story correctly? (13:45) YES / NO 19. What was added to the code? (15:15) - _________________ were applied - Each image was _____________________ - Each image was _________________ to 28 x 28 pixels. 20. Does the program work after the changes were made? (16:05) YES / NO Direct Episode Link: https://www.youtube.com/watch?v=6nGCGYWMObE&t=181s 14 © 2019, K. Maxson  Episode Guides Without Timestamps https://www.teacherspayteachers.com/Store/Karen-Maxson  Name: ________________________ Crash Course Artificial Intelligence Episode #1 : What is Artificial Intelligence? 1. Which of the following use A.I.? (Circle all that apply) A. Banks C. YouTube B. Doctors D. Apps for cell phones 2. How do you feel about A.I.? ________________________________________________ ____________________________________________________________________ ____________________________________________________________________ 3. “Writers like to imagine a more _________________________ A.I.” 4. This is a realistic way to think of A.I. TRUE / FALSE 5. “A machine is said to have artificial intelligence if it can _____________________________, potentially __________________ from the data, and use that knowledge to _____________ and achieve specific goals.” 6. Today’s A.I. still can’t do most of the things that humans do. TRUE / FALSE 7. A.I. is used to screen ____________________ applications. Do you think this is fair? Explain. ____________________________________________________________________ ____________________________________________________________________ ____________________________________________________________________ 8. When was the term “artificial intelligence” coined? _________________________________ 16 © 2019, K. Maxson  9. What important conference is responsible for a lot of advancements in A.I.? ____________________________________________________________________ 10. A.I. has developed FASTER / SLOWER than these early thinkers predicted. 11. When did the “A.I. Winter” end? _____________________________________________ 12. What “two big developments” helped to spur the A.I. revolution forward? 1. Increase in ____________________________ 2. __________________ and ________________________ 13. What increases the speed of a computer? A. More hard drives C. Faster internet speed B. More circuit boards D. More transistors 14. When was the first iPhone released? A. 2001 C. 2007 B. 2005 D. 2011 Direct Episode Link: https://www.youtube.com/watch?v=a0_lo_GDcFw 17 © 2019, K. Maxson  Name: ________________________ Crash Course Artificial Intelligence Episode #2: Supervised Learning 1. What process helps computers learn? _________________________________________ 2. What are the “three main types of learning?” ____________________________________________________________________ 3. “Supervised learning is the process of learning with _______________________________.” 4. What two things does an A.I. need to learn? A. Humans to supervise C. Answer key B. Computing power D. Data 5. Scientists were inspired by the human brain when they made some kinds of A.I. TRUE / FALSE 6. What are the three basic parts of a neuron? ____________________________________________________________________ 7. What machine did Rosenblatt build? What was he trying to teach it to do? ____________________________________________________________________ 8. What special weight represents the threshold? ____________________________________ 9. Generalized A.I. was successfully created in the early 2010’s. TRUE / FALSE 10. Name 2 different activation functions. _________________________________________ 11. At the beginning of supervised learning, the A.I. makes random guesses. TRUE / FALSE 18 © 2019, K. Maxson  12. When John Green Bot is correct, the update rule will add “__________ to the weight.” 13. John Green Bot learns from SUCCESS / FAILURE. 14. You can put the results of a supervised learning test into “a table, called a _________________ ______________________.” 15. “____________________ tells us how much you should _____________ your program when it says it’s found something.” 16. “____________________ tells you how much your program can find of the thing you’re looking for.” 17. What is the “key to most A.I. challenges? _______________________________________ Direct Episode Link: https://www.youtube.com/watch?v=4qVRBYAdLAo 19 © 2019, K. Maxson  Name: ________________________ Crash Course Artificial Intelligence Episode #3: Neural Networks and Deep Learning 1. How many neurons does our brain have? A. 100 thousand C. 100 billion B. 100 million D. 100 trillion 2. “Not that long ago, a big challenge in AI was ____________________________ recognition.” 3. Who helped to develop AI that could recognize images? _____________________________ 4. What was their international dataset called? How many images does it have? ____________________________________________________________________ 5. What made AlexNet different? ______________________________________________ 6. “All neural networks are made up of an ______________ layer, an ________________ layer, and any number of _______________ layers in between.” 7. Each pixel can be represented with 3 numbers that correspond to the amount of ____________, _______________, and ______________ in the pixel. 8. What layer is “where the final hidden layer outputs are mathematically combined to answer the problem?” ____________________________________________________________________ 9. What is “the key to neural networks and…all of A.I.?” _______________________________ 10. Every hidden neuron looks for all of the components needed to determine if the picture is what they are looking for. (i.e. a dog, or cat) TRUE / FALSE 11. How many neurons are in the output layer? __________________________ 12. Deeper neural networks “are networks with more _________________________ to do deep _________________.” 13. What is one way that neural networks are used in your life already? _____________________ ____________________________________________________________________ Direct Episode Link: 20 https://www.youtube.com/watch?v=oV3ZY6tJiA0 © 2019, K. Maxson  Name: ________________________ Crash Course Artificial Intelligence Episode #4: Training Neural Networks 1. Neural networks are now big enough that some can do tasks better than humans. TRUE / FALSE 2. What algorithm do neural networks use to handle mistakes? __________________________ 3. What are the “two main parts” of neural networks? _________________________________ 4. “The task of finding the best weights for a neural network architecture is called __________.” A. Backpropagation C. Nodular B. Optimization D. Synapse 5. What is “the goal of linear regression?” _________________________________________ ____________________________________________________________________ 6. “By connecting together many simple _______________ with __________________, a neural network can learn to solve complicated _____________________.” 7. What do neurons multiply the given features by? __________________________________ 8. What represents the error when “the predicted answer and the correct answer is more than just one number?” ____________________________________________________________________ 9. What algorithm did scientists create “to help neural networks learn?” ____________________________________________________________________ 10. A _____________________________ is “where the weights make the error relatively small, but not the smallest it could be.” 11. Ideally, you should explore different solutions at the same time, on the same neural network. However, that just isn’t feasible yet. TRUE / FALSE 12. “Over time, backpropagation will adjust the neuron’s _________________, so that the neural networks output matches the ________________________...That’s called _____________ _________________________________________.” Direct Episode Link: 21 https://www.youtube.com/watch?v=lgKrup5oi_A © 2019, K. Maxson  Name: ________________________ Crash Course Artificial Intelligence Episode #5: how to Make an AI read your handwriting (LAB) 1. In this episode “we’ll tackle a ________________ together and program a ___________________________ to recognize __________________________.” 2. What language will be used to write the code in this episode? A. JavaScript C. C# B. Python D. PHP 3. Where can you find the code that the episode is working on today? ____________________________________________________________________ 4. “Our goal is to program a neural network to recognize _____________________ letters and convert them to ______________________________.” 5. “Usually with a project like this, we’d have to write code to figure out where one letter ends and another begins...That’s called the _______________________________________.” 6. Step 1: “Find or create a ______________________________ to train our neural network.” 7. Step 2: “Create a ________________________________.” 8. Step 3: “Train, _________________, and tweak our code until we feel that it’s ___________________ enough.” Step 4: Scan John-Green-Bot’s handwritten pages and use our newly trained neural network to convert them into typed text!” 9. What dataset (abbreviated) will be used for this lab? ________________________________ 10. Step 1.2 includes training the network with 100,000 images. TRUE / FALSE 22 © 2019, K. Maxson  11. “___________________________________ is a very important step to training a good neural network.” 12. What type of neural network will they be using in their episode? ____________________________________________________________________ 13. “The size of our output layer depends on the number of __________________________ that we want our neural network to guess.” 14. How many output neurons will this network have? A. 26 C. 130 B. 52 D. 1,300 15. What number does the program print for every epoch? ____________________________________________________________________ You want to see this number go UP / DOWN over time. 16. What is the tradeoff if you add more layers and neurons to your network to make it more accurate? ____________________________________________________________________ 17. Where are the images for this step being stored? __________________________________ 18. Does the program process the story correctly? YES / NO 19. What was added to the code? - _________________ were applied - Each image was _____________________ - Each image was _________________ to 28 x 28 pixels. 20. Does the program work after the changes were made? YES / NO Direct Episode Link: https://www.youtube.com/watch?v=6nGCGYWMObE&t=181s 23 © 2019, K. Maxson Answer Keys https://www.teacherspayteachers.com/Store/Karen-Maxson  Crash Course Artificial Intelligence Episode #1 : What is Artificial Intelligence? 1. Which of the following use A.I.? (0:10) (Circle all that apply) A. Banks C. YouTube B. Doctors D. Apps for cell phones 2. How do you feel about A.I.? (0:30) Answers may vary 3. “Writers like to imagine a more generalized A.I.” (1:30) 4. This is a realistic way to think of A.I. (1:35) FALSE 5. “A machine is said to have artificial intelligence if it can interpret data , potentially learn from the data, and use that knowledge to adapt_ and achieve specific goals.” (1:50) 6. Today’s A.I. still can’t do most of the things that humans do. (3:38) TRUE 7. A.I. is used to screen college/job applications. (4:21) Do you think this is fair? Explain. Answers may vary 8. When was the term “artificial intelligence” coined? (4:55) 1956 9. What important conference is responsible for a lot of advancements in A.I.? (5:00) Dartmouth Summer Research Project 10. A.I. has developed SLOWER than these early thinkers predicted. (5:47) 11. When did the “A.I. Winter” end? (7:15) Approximately 2010 12. What “two big developments” helped to spur the A.I. revolution forward? 1. Increase in computing power (7:43) 2. Internet and Social Media (9:50) 13. What increases the speed of a computer? (8:35) D. More transistors 14. When was the first iPhone released? (8:48) C. 2007 Direct Episode Link: https://www.youtube.com/watch?v=a0_lo_GDcFw 25 © 2019, K. Maxson  Crash Course Artificial Intelligence Episode #2: Supervised Learning 1. What process helps computers learn? (0:30) Supervised Learning 2. What are the “three main types of learning?” (1:03) Supervised, Unsupervised, and Reinforcement 3. “Supervised learning is the process of learning with training labels.” (1:35) 4. What two things does an A.I. need to learn? (2:05) B. Computing power D. Data 5. Scientists were inspired by the human brain when they made some kinds of A.I. (2:45) TRUE 6. What are the three basic parts of a neuron? (2:55) Dendrites, cell body, and axon 7. What machine did Rosenblatt build? What was he trying to teach it to do? (3:45) - The Perception - He was trying to teach it to classify shapes as triangles, or “not triangles”. 8. What special weight represents the threshold? (5:35) The Bias 9. Generalized A.I. was successfully created in the early 2010’s. (6:40) FALSE 10. Name 2 different activation functions. (7:25) Students choose two of the following: Logistic, Linear, Unit Step, Hyperbolic Tangent, Piecewise Linear, or Sign. 11. At the beginning of supervised learning, the A.I. makes random guesses. (8:58) TRUE 12. When John Green Bot is correct, the update rule will add “ 0 to the weight.” (9:25) 13. John Green Bot learns from FAILURE. (9:45) 14. You can put the results of a supervised learning test into “a table, called a _Confusion Matrix_.” (11:50) 15. “ Precision tells us how much you should trust your program when it says it’s found something.” (12:25) 16. “ Recall tells you how much your program can find of the thing you’re looking for.” (12:48) 17. What is the “key to most A.I. challenges? (13:53) “Figuring out what criteria to use.” Direct Episode Link: https://www.youtube.com/watch?v=4qVRBYAdLAo 26 © 2019, K. Maxson  Crash Course Artificial Intelligence Episode #3: Neural Networks and Deep Learning 1. How many neurons does our brain have? (0:15) C. 100 billion 2. “Not that long ago, a big challenge in AI was real-world image recognition.” (0:52) 3. Who helped to develop AI that could recognize images? (1:24) Fei-Fei Li 4. What was their international dataset called? How many images does it have? (1:43) ImageNet 3.2 million 5. What made AlexNet different? (2:40) “He used a lot of hidden layers” and “faster computation hardware.” 6. “All neural networks are made up of an input layer, an output layer, and any number of hidden layers in between.” (3:24) 7. Each pixel can be represented with 3 numbers that correspond to the amount of red , blue , and green in the pixel. (4:22) 8. What layer is “where the final hidden layer outputs are mathematically combined to answer the problem?” (5:12) Output layer 9. What is “the key to neural networks and…all of A.I.?” (5:47) Math 10. Every hidden neuron looks for all of the components needed to determine if the picture is what they are looking for. (i.e. a dog, or cat) (7:30) FALSE 11. How many neurons are in the output layer? (8:34) One 12. Deeper neural networks “are networks with more hidden layers to do deep learning.” (9:42) 13. What is one way that neural networks are used in your life already? (10:45) Answers may vary Direct Episode Link: https://www.youtube.com/watch?v=oV3ZY6tJiA0 27 © 2019, K. Maxson  Crash Course Artificial Intelligence Episode #4: Training Neural Networks 1. Neural networks are now big enough that some can do tasks better than humans. (0:12) TRUE 2. What algorithm do neural networks use to handle mistakes? (0:40) Backpropagation 3. What are the “two main parts” of neural networks? (0:50) “The architecture and the weights” 4. “The task of finding the best weights for a neural network architecture is called ______.” (1:12) B. Optimization 5. What is “the goal of linear regression?” (2:12) The goal is “to adjust the line to make the error as small as possible.” 6. “By connecting together many simple neurons with weights , a neural network can learn to solve complicated problems.” (3:10) 7. What do neurons multiply the given features by? (4:30) They multiply them by the weights each has been assigned. 8. What represents the error when “the predicted answer and the correct answer is more than just one number?” (5:10) A Loss Function 9. What algorithm did scientists create “to help neural networks learn?” (5:42) Backpropagation of the Error 10. A local optimal solution is “where the weights make the error relatively small, but not the smallest it could be.” (8:18) 11. Ideally, you should explore different solutions at the same time, on the same neural network. However, that just isn’t feasible yet. (9:10) FALSE 12. “Over time, backpropagation will adjust the neuron’s weights , so that the neural networks output matches the training data...That’s called fitting to the training data.” (10:38) Direct Episode Link: https://www.youtube.com/watch?v=lgKrup5oi_A 28 © 2019, K. Maxson  Crash Course Artificial Intelligence Episode #5: how to Make an AI read your handwriting (LAB) 1. In this episode “we’ll tackle a projects together and program a neural network to recognize handwritten letters.” (1:15) 2. What language will be used to write the code in this episode? (1:30) B. Python 3. Where can you find the code that the episode is working on today? (1:35) There is a link to it in the video’s description. 4. “Our goal is to program a neural network to recognize handwritten letters and convert them to typed text.” (2:10) 5. “Usually with a project like this, we’d have to write code to figure out where one letter ends and another begins...That’s called the segmentation.” (2:25) 6. Step 1: “Find or create a labeled dataset to train our neural network.” (3:17) 7. Step 2: “Create a neural network.” (3:35) 8. Step 3: “Train, test , and tweak our code until we feel that it’s accurate enough.” (3:48) Step 4: Scan John-Green-Bot’s handwritten pages and use our newly trained neural network to convert them into typed text!” 9. What dataset (abbreviated) will be used for this lab? (4:15) EMNIST 10. Step 1.2 includes training the network with 100,000 images. (5:00) FALSE 11. “ Data collection is a very important step to training a good neural network.” (5:57) 12. What type of neural network will they be using in their episode? (6:50) They are using a “multi-layer perceptron neural network” (MLP) 13. “The size of our output layer depends on the number of label types that we want our neural network to guess.” (7:43) 29 © 2019, K. Maxson  14. How many output neurons will this network have? (7:50) A. 26 15. What number does the program print for every epoch? (9:50) Error of the Loss Function You want to see this number go DOWN over time. 16. What is the tradeoff if you add more layers and neurons to your network to make it more accurate? (11:20) It will run slower. 17. Where are the images for this step being stored? (12:47) A GitHub Repository 18. Does the program process the story correctly? (13:45) NO 19. What was added to the code? (15:15) - Filters were applied - Each image was centered_ - Each image was resized to 28 x 28 pixels. 20. Does the program work after the changes were made? (16:05) YES (mostly) Direct Episode Link: https://www.youtube.com/watch?v=6nGCGYWMObE&t=181s 30 © 2019, K. Maxson