![Building a basic DL Model using python
import numpy as np
import tensorflow as tf
coefficients = np.array([[1], [-20], [25]])
w = tf.Variable([0],dtype=tf.float32)
x = tf.placeholder(tf.float32, [3,1])
cost = x[0][0]*w**2 + x[1][0]*w + x[2][0] # (w-5)**2
train = tf.train.GradientDescentOptimizer(0.01).minimize(cost)
init = tf.global_variables_initializer()
with tf.Session() as session:
session.run(init)
print(session.run(w))
for i in range(1000):
session.run(train, feed_dict={x:coefficients})
print(session.run(w))](https://image.slidesharecdn.com/presentationv3-181114040738/85/Presentation-v3-30-320.jpg)
More Related Content
Similar to Presentation v3 (20)
![[Srijan Wednesday Webinars] Artificial Intelligence & the Future of Business](https://cdn.slidesharecdn.com/ss_thumbnails/disrupt4-171207140409-thumbnail.jpg?width=560&fit=bounds)
![[DSC Europe 22] On the Aspects of Artificial Intelligence and Robotic Autonom...](https://cdn.slidesharecdn.com/ss_thumbnails/pushandutta-artificialintelligenceinrobotautonomy-221130084927-d0f4c0c6-thumbnail.jpg?width=560&fit=bounds)
Presentation v3
- 1. ARTIFICIAL INTELLIGENCE AND IoT From concept to applications Dr. Muhammad Al-Qurishi Data Scientist, Research Chair of Pervasive and Mobile Computing (CPMC)
- 2. What is AI?
- 3. “the science and engineering of making intelligent machines.” John McCarthy, one of the godfathers of AI
- 4. AI also A branch of computer science dealing with the simulation of intelligent behavior in computers The capability of a machine to imitate intelligent human behavior A computer system able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision- making, and translation between languages
- 5. AI can be a pile of if-then statements or a complex statistical model mapping raw sensory data to symbolic categories The if-then statements are simply rules explicitly programmed by a human hand. Taken together, these if-then statements are sometimes called rules engines, expert systems, knowledge graphs or symbolic AI Symbolic AI
- 6. Symbolic AI Good, Old-Fashioned AI (AI) Symbolic Reasoning (Symbolic AI) is one of the AI branches Example of the intelligence they mimic An accountant with knowledge of the tax code, who takes information you feed it, runs the information through a set of static rules, and gives your the amount of taxes you owe as a result.
- 7. uses Signs and Symbols Signifiers and Signifieds Symbolic AI Combinations of symbols that express their interrelations called reasoning. • All men are mortal; Caius is a man; therefore Caius is mortal. CredFinder a simple rule- based system, by Dr. Majed Al-rubaian et. al ASONAM 2016
- 8. Is it really true? A wag would say that true AI is whatever computers can’t do yet
- 9. “the science and engineering of making intelligent machines.” John McCarthy, one of the godfathers of AI AI Pioneer and Deep learning Heroes “Artificial Intelligence Will Almost Certainly Surpass Humans" Geoffrey Hinton the father of deep learning “Computer are not just going to replace manual labor but also mental labor" Yoshua Bengio, AI Pioneer “AI is the new electricity" Andrew Ng the founder of deeplearning.ai “AI is the real science and deep learning is the way to go in the future" Ian Goodfelow the founder of GANs
- 10. Machine Learning A field of study that gives computers the ability to learn without being explicitly programmed
- 11. Machine Learning Feature Extraction A set of rules Dog Bird Cat Fish
- 12. Machine Learning Feature Extraction A Machine learning model Dog Bird Cat Fish
- 13. Real-life Machine learning Examples Recommender systems • Amazon recommender systems • Social networks recommender systems Banks • approving a loan application Telecommunication Companies • Prediction of customer un/ subcribe • Chatbot, using face recognition to logging into our phones or even computer games And many others
- 14. ML techniques and algorithms Supervised learning Unsupervised learning Machine Learning
- 15. • How do we supervise a machine learning model? • Training it with labeled dataset • What does a labeled dataset look like? Supervised Classification • The process of predicting a discrete class label, or category Regression • The process of predicting a continuous value as opposed to predicting a categorical value in classification
- 16. • What do you think unsupervised learning means? • Exactly as it sounds • Trains on the dataset • Draws conclusions on unlabeled data Unsupervised Unsupervised machine learning techniques Dimension reduction market basket analysis Clustering Clustering discovering structure Summarization anomaly detection
- 17. Supervised vs. Unsupervised • Classification • Classifies labeled data • Regression • Predict trends using previous labeled data • Has more evaluation techniques than unsupervised • Controlled environment Supervised • Clustering • Finds patterns and groupings from unlabeled data • Has fewer evaluation techniques than supervised • Less controlled environment Unsupervised
- 18. Example of Supervised and Unsupervised • Naïve Bayes • Decision Tree • Support Vector Machine • Random Forest • Leaner Regression • Neural Network • K-nearst Neighbor Supervised • K-means • Principle Component Analysis Unsupervised
- 19. Building a basic ML Model using python
- 20. Python ML Frameworks and Libraries
- 21. Basic ML Example using Scikit-learn
- 22. Artificial Intelligence (AI) vs. Machine Learning vs. Deep Learning and also Deep symbolic learning Artificial Intelligence Machine learning Deep learning
- 23. Why Now? Scale drives deep learning progress
- 24. Deep learning Being able to train a big enough NN Two things have to be considered to get high level of performance Huge amount of labeled data eventually you run out of data That only works up to a point because or eventually then your network is so big that it takes too long to train
- 25. Deep learning It is important to have a fast computation unit Idea CodeExperiment
- 26. Example of DNN training time
- 27. Example of simple NN Linear neurons
- 28. Activation functions Sigmoid TanH Rectified linear unit (ReLU) leaky rectified linear unit (RReLU)
- 29. Deep learning Frameworks and Libraries CNTK Lasagne PaddlePaddle • Ease of programming (development and deployment) • Running speed • Truly open (open source with good governance) Choosing deep learning frameworks
- 30. Building a basic DL Model using python import numpy as np import tensorflow as tf coefficients = np.array([[1], [-20], [25]]) w = tf.Variable([0],dtype=tf.float32) x = tf.placeholder(tf.float32, [3,1]) cost = x[0][0]*w**2 + x[1][0]*w + x[2][0] # (w-5)**2 train = tf.train.GradientDescentOptimizer(0.01).minimize(cost) init = tf.global_variables_initializer() with tf.Session() as session: session.run(init) print(session.run(w)) for i in range(1000): session.run(train, feed_dict={x:coefficients}) print(session.run(w))
- 31. Deep learning resources • Book : Deep learning by Ian Goodfelow et. al • Deeplearning.ai • FastAI • OpenAI • DeepMind • FacebookAI • Google Brain
- 32. Symbolic AI vs. Deep learning • The main drawback is the difficulty of revising beliefs once they were encoded in a rules engine • Expert systems are monotonic; • The more rules you add, the more knowledge is encoded in the system, but additional rules can’t undo old knowledge • Open research line: • Can we fuse the ability of deep neural nets to learn probabilistic correlations from scratch alongside abstract and higher-order concepts, which are useful in compressing data and combining it in new ways? • Deep symbolic learning • Enabling deep neural networks to manipulate, generate and otherwise cohabitate with concepts expressed in strings of characters, could help solve explainability, because, after all, humans communicate with signs and symbols, and that is what we desire from machines
- 33. Symbolic AI vs. Deep learning • Deep learning biggest weaknesses are • Model Explanation • Data hungry • Geoff Hinton himself has expressed skepticism about whether backpropagation, the workhorse of deep neural nets, will be the way forward for AI. • Open research: • Hyper-parameters • Work on unsupervised learning (learning from unlabeled data) will lead to our next breakthroughs. • GANs and CycleGANs
- 35. What is the IoT?
- 36. Typical shopping
- 37. IoT-based shopping Search patterns: (Time distance/traffic..) One or more parameter may affected by other parameters. e.g. Cart picks the optimal path to choose the proper items on shelves Delivery
- 38. IoT device is no a Computer • Special purpose thing • Can be good for one task or multi-task that all related to one thing, but terrible for another • such as audio player is great for playing but not good in playing video • Computer can do both • But less efficiently (Froe example camera of the laptop)
- 39. What Support IoT • Cost of Hardware and software are cheaper than before • You can get a laptop 1500SAR • IoT will be cheaper, the computational ability of it is a lot less than laptops • You can add IoT to old machines with low cost (example, car) • Hardware size • IoT is small and light • Computational technology has also adaptable to that • Speed of the computational technology • Speech-to-Text, audio processing and network communication
- 40. What Support IoT • Internet Access • Internet access has improved • Internet is reachable from a lot of places • Wi-Fi • Cheaper than wired • Makes IoT technology much more usable and much more possible • Data usage it not that expensive • High bandwidth • Transmit lots of data fast (YouTube videos)
- 41. Pervasive and Cloud computing • Pervasive nature of IoT • IoT is not only everywhere, but also networked • Can be an interface to the cloud (example refrigerator search) • A window to massive database and some massive computational resource • The actual computation is the one send to the cloud • The cloud process it and give the result back • You can think of many examples • IoT can work as an access points for something much bigger
- 42. IoT Privacy-and-security • Depending on Technology and Infrastructure • Big risk if something goes down • power outages • Network outages • Blackout • IoT devices are observing us: • Location (reporting you GPS) • Health monitoring (store) • Media system • Purchasing systems
- 43. IoT frameworks and library https://phodal.github.io/awesome-iot/
- 44. Thank You