Artificial Intelligence Machine Learning and Deep Learning Updated

Artificial Intelligence, Machine Learning, and Deep Learning Updated on: 18 th May, 2020 By: Celeste Ng Direct quote from sources: As indicated in each individual slide

Artificial Intelligence AI is a technique which enables machine to mimic human behaviors and mind (AI是一種技術使機器 能�模仿人類行為和 思維) Source: https: //www. youtube. com/w atch? v=WSbgixd. C 9 g 8&t=19 s

Machine Learning Machine learning is • a subset of AI technique, which uses statistical methods to enable machines to improve with experience (機器學習是 AI技術的一個子集，它使用統計 方法使機器能�通過經驗進行改 進) Machine learning is • designed in a way that it can learn and improve over time, when exposes to new data Source: https: //www. youtube. com/watch? v=WSbgixd. C 9 g 8&t=19 s

Deep Learning Deep learning is • a particular kind of ML that is inspired by the functionality of brain cells (neuron), which leads to the concept of artificial neural network (ANN) (一種特殊的機器學習, 來自腦細胞神經元功能的� 神經元 發，這導致了人 神經網絡的概念) Its Learning can be supervised, semisupervised or unsupervised. --- Source: Wikipedia, 2020 Source: https: //www. youtube. com/watch? v=WSbgixd. C 9 g 8&t=19 s

AI vs. Machine Learning vs. Deep Learning - Definition It is inspired by structure of a human brain. Source: https: //towardsdatascience. com/cousins-of-artificial-intelligence-dda 4 edc 27 b 55

AI vs. Machine Learning vs. Deep Learning - timeline Source:

Types of Machine Learning Technique – nature & objective 識別集群 unsupervised learning: is to find similarities and differences between differences Source: data points https: //towardsdatascience. com/reinfo supervised learning: is to find a correct action for action performing a task rcement-learning-101 -e 24 b 50 e 1 d 292 reinforcement learning: is to find a suitable action model suitable action that would maximize the total cumulative reward of the agent 尋找合適的決定模型，以最大 化Agent的總累積獎勵

Types of Machine Learning Technique – requirements Source: https: //www. analyticsvidhya. com/machine-learning/types_of_ml/

Types of Machine Learning Technique – overall key points Source: https: //www. newtechdojo. com/list-machine-learning-algorithms/

Types of Machine Learning Technique – types of learning, algorithm & application areas Source: https: //www. codeproject. com/Articles/5245488/Introduction-to-Machine. Learning-and-ML-NET-Part-1

Supervised Machine Learning Types: Algorithms: Source: https: //vas 3 k. com/blog/machine_learning/? ref=hn

Unsupervised Machine Learning Types: Algorithms: Source: https: //vas 3 k. com/blog/machine_learning/? ref=hn

Supervised vs. Unsupervised Machine Learning Source: https: //www. researchgate. net/figure/Supervised-learning-and-unsupervisedlearning-Supervised-learning-uses-annotation_fig 1_329533120

Supervised vs. Unsupervised Learning 因變量和一個或多個自變量 之間的線性關係 Source: https: //www. researchgate. net/figure/Examples-of-Supervised-Learning-Linear. Regression-and-Unsupervised-Learning_fig 3_336550812

Example of Supervised Learning Technique Equation has the form Y= a + b. X Source: https: //towardsdatascience. com/simple-machine-learning-model-in-python-in-5 -linesof-code-fe 03 d 72 e 78 c 6

Example of Supervised Learning Technique Source: https: //towardsdatascience. com/simple-machine-learning-model-in-python-in-5 -linesof-code-fe 03 d 72 e 78 c 6

Machine Learning Applications • 1. Virtual Personal Assistants – Siri, Alexa, Google Now are some of the popular examples of virtual personal assistants. As the name suggests, they assist in finding information, when information asked over voice. • 2. Predictions while Commuting – Traffic Predictions: We all have been using GPS navigation services. While we do that, our current locations and velocities are being saved at a central server for managing traffic. – Uber use ML to define price surge hours by predicting the rider demand Source: https: //medium. com/app-affairs/9 -applications-of-machine-learning-from-day-today-life-112 a 47 a 429 d 0

Machine Learning Applications • 3. Videos Surveillance – A difficult job to do and boring as well. This is why the idea of training computers to do this job makes sense. – It is possible to detect crime before they happen. They track unusual happen behaviour of people like standing motionless for a long time, stumbling, or napping on benches etc. • 4. Social Media Services – Personalizing your news feed (動態消息), providing better ads targeting – The applications of ML. • People You May Know: Machine learning works on a simple concept: understanding with experiences. Facebook continuously notices the friends that you connect with, the profiles that you visit very often, your interests, workplace, or a group that you share with someone • Face Recognition: You upload a picture of you with a friend and Facebook instantly recognizes that friend. Source: https: //medium. com/app-affairs/9 -applications-of-machine-learning-from-day-today-life-112 a 47 a 429 d 0

Machine Learning Applications • 5. Email Spam and Malware Filtering – Over 325, 000 malwares (惡意軟件) are detected everyday and each piece of code is 90– 98% similar to its previous versions. – The system security programs that are powered by machine learning understand the coding pattern. Therefore, they detects new malware with 2– 10% variation easily and offer protection against them. • 6. Online Customer Support – Chatbots tend to extract information from the website and present it to the customers – Chatbots advances with time. They tend to understand the user queries better and serve them with better answers, which is possible due to its machine learning algorithms. Source: https: //medium. com/app-affairs/9 -applications-of-machine-learning-from-day-today-life-112 a 47 a 429 d 0

Machine Learning Applications • 7. Search Engine Result Refining – Google and other search engines use machine learning to improve the search results for you. – Every time you execute a search, the algorithms at the backend keep a watch at how you respond to the results. If you open the top results and stay on results the web page for long, the search engine assumes that the results it displayed were in accordance to the query. Source: https: //medium. com/app-affairs/9 -applications-of-machine-learning-from-day-today-life-112 a 47 a 429 d 0

Machine Learning Applications • 8. Product Recommendations – You shopped for a product online few days back and then you keep receiving emails for shopping suggestions. – If not this, then you might have noticed that the shopping website or the app recommends you some items that somehow matches with your taste. Certainly, this refines the shopping taste experience Source: https: //medium. com/app-affairs/9 -applications-of-machine-learning-from-day-today-life-112 a 47 a 429 d 0

Machine Learning Applications • 9. Online Fraud Detection – Machine learning is proving its potential to make cyberspace a secure place and tracking monetary frauds online is one of its examples. – For example: Paypal is using ML for protection against money laundering. The company uses a set of tools that helps them to compare millions of transactions taking place and distinguish between legitimate or illegitimate transactions taking place between the buyers and sellers – Source: https: //medium. com/app-affairs/9 -applications-of-machine-learning-from-day-today-life-112 a 47 a 429 d 0

Machine Learning Applications - overview Source: https: //www. oreilly. com/library /view/java-deeplearning/9781788997454/a 8 fce 962 -51 dd-4 e 29 -a 7 f 99 bf 4 fd 245 b 1 d. xhtml

Deep Learning (appeared earlier) Deep learning is • a particular kind of ML that is inspired by the functionality of brain cells (neuron), which leads to the concept of artificial neural network (ANN) (一種特殊的機器學習, 來自腦細胞神經元功能的� 神經元 發，這導致了人 神經網絡的概念) Its Learning can be supervised, semisupervised or unsupervised. --- Source: Wikipedia, 2020 …and reinforcement learning (as in Alpha. Go Zero) --- Source: https: //dl. acm. org/doi/pdf/10. 1145/3206157. 3206174; https: //web. stanford. edu/~surag/posts/alphazero. html Source: https: //www. youtube. com/watch? v=WSbgixd. C 9 g 8&t=19 s

Deep Learning Deep learning is • ANN is modeled using layers of artificial neurons or computational units to receive input and apply an activation function along with threshold. • In simple model the first layer is input layer, followed by a hidden layer, and lastly by an output layer. Each layer contains one or more neurons. Source: https: //towardsdatascience. com/cousins-of-artificial-intelligence-dda 4 edc 27 b 55

Deep Learning – a simple example How you recognize square from other shapes? • First thing we do is check whether the figure has four lines. • If yes, we further check if all are lines are connected and closed. • If yes we finally check if all are perpendicular(垂直的) and all sides are equal. We consider the figure as square if it satisfies all the conditions. • As we saw in the example as nested hierarchy of concepts. • So we took a complex task of identifying a square and broken down into simpler tasks. • Deep learning also does the same thing but at a larger scale (更大的規模). learning Source: https: //towardsdatascience. com/cousins-of-artificial-intelligence-dda 4 edc 27 b 55

Machine learning vs Deep Learning Machine learning – human feature-engineering Deep learning – featureidentification without human intervention Source: https: //www. youtube. com/watch? v=6 M 5 VXKLf 4 D 4

Deep Learning – a real example Source: https: //www. youtube. com/watch? v=6 M 5 VXKLf 4 D 4

Deep Learning – a real example: training algorithm to identify hand-written digit Input layer, output layer & hidden layer Each pixel is fed into a neuron this forms the “ Input Layer” Layer Each image is represented by 28 x 28 pixels Source: https: //www. youtube. com/watch? v=6 M 5 VXKLf 4 D 4 Each neuron represents a digit Hidden Layer

Deep Learning – a real example Connected channels, weighted channels, Information is transferred from one layer to another through “ connected channels ” Source: https: //www. youtube. com/watch? v=6 M 5 VXKLf 4 D 4 Each channel has a value attached to it, thus, it is called “ weighted channels ”

Deep Learning – a real example Bias, weighted sum of inputs, activation function Each neuron has a unique number associated with it, called “ bias ” Source: https: //www. youtube. com/watch? v=6 M 5 VXKLf 4 D 4 “Bias ” is added to the “ weighted sum of inputs ” reaching the neuron, which is then applied to a function, known as “ activation function ” Bias Weighted sum of inputs

Deep Learning – a real example Activation function results The results of the “ activation function ” determines which neurons are activated. The activated neurons passed the information to the following layers. … continue until the 2 nd last layer Source: https: //www. youtube. com/watch? v=6 M 5 VXKLf 4 D 4

Deep Learning – a real example Final result The one neuron activated in the output layer corresponds to the input digit Source: https: //www. youtube. com/watch? v=6 M 5 VXKLf 4 D 4

Deep Learning – a real example Adjustment The weights and bias are continuously adjusted to produce a “well-trained network” network Source: https: //www. youtube. com/watch? v=6 M 5 VXKLf 4 D 4

Deep Learning – other examples Source: https: //www. youtube. com/watch? v=6 M 5 VXKLf 4 D 4

Deep Learning – unique requirements 1. Deep learning – data dependency: deep learning efficiency depends on massive amount of data. Source: https: //www. youtube. com/watch? v=WSbgixd. C 9 g 8&t=19 s

Deep Learning – unique requirements 2. Deep learning – hardware dependency: deep learning relies on high end hardware processing power of GPUs Source: https: //www. youtube. com/watch? v=WSbgixd. C 9 g 8&t=19 s

Deep Learning – unique requirements 3. Amount of time required to train the algorithm ranges from hours to months. Source: https: //www. youtube. com/watch? v=6 M 5 VXKLf 4 D 4

Examples of Deep Learning Code Source: https: //chatbotslife. com/deep-learning-in-7 -lines-of-code-7879 a 8 ef 8 cfb

Examples of Deep Learning Code Source: https: //chatbotslife. com/deep-learning-in-7 -lines-of-code-7879 a 8 ef 8 cfb

Deep Learning Frameworks Source: https: //www. youtube. com/watch? v=_FBXfar. XKu. A

Deep Learning Frameworks Source: https: //www. youtube. com/watch? v=_FBXfar. XKu. A

Deep Learning Frameworks Source: https: //www. youtube. com/watch? v=_FBXfar. XKu. A

Deep Learning Frameworks Source: https: //www. youtube. com/watch? v=_FBXfar. XKu. A

Deep Learning Frameworks Source: https: //www. youtube. com/watch? v=_FBXfar. XKu. A

Deep Learning Frameworks Source: https: //www. youtube. com/watch? v=_FBXfar. XKu. A

Deep Learning Frameworks Source: https: //www. youtube. com/watch? v=_FBXfar. XKu. A

Deep Learning Frameworks Source: https: //www. youtube. com/watch? v=_FBXfar. XKu. A

Deep Learning Frameworks Source: https: //www. youtube. com/watch? v=_FBXfar. XKu. A

Deep Learning Frameworks Source: https: //www. youtube. com/watch? v=_FBXfar. XKu. A

Other references 1. https: //www. datarobot. com/wiki/model/ 2. https: //www. analyticsvidhya. com/blog/2017/09 /common-machine-learning-algorithms/