Artificial Neural Networks: Pointers
Download as PPTX, PDF1 like1,248 views
The document discusses the fundamentals of artificial neural networks (ANNs) by comparing them to biological neural networks, highlighting key aspects such as neuron structure, learning processes, and the significance of activation functions. It explains the McCulloch-Pitts processing unit and introduces techniques like backpropagation and gradient descent for training multi-layered networks. Various applications of neural networks, including face recognition and self-driving cars, are also mentioned.
Artificial Neural Networks: Pointers
- 1. Artificial Neural Networks Fariz Darari (fariz@ui.ac.id) Feel free to contact me for any partnership opportunities!
- 2. Organic Neural Network ā¢ The human brain has about 1011 neurons ā¢ Switching time 0.001s (computer ā 10-10s) ā¢ Connections per neuron: 104 - 105 ā¢ 0.1s for face recognition! ā¢ Strengths: Parallelism and distributedness
- 3. Biological Neurons ā¢ Dendrit menerima input informasi dalam bentuk sinyal elektrik, yang diakumulasikan di badan sel saraf (= soma). ā¢ Ketika akumulasi informasi mencapai threshold tertentu, sel saraf akan menembakkan output informasi yang dikirim melalui axon ā¢ Axon tersambung dengan dendrit yang ada di badan sel saraf lain melalui synapses ā¢ Pemelajaran/learning dilakukan dengan adaptasi synaptical weight
- 4. McCulloch-Pitts Processing Unit (1943) Contoh:
- 5. ANN: Basic Idea ā¢ Artificial Neuron ā¢ Each input is multiplied by a weighting factor ā¢ Output is: 1 if sum of weighted inputs exceeds threshold; 0 otherwise ā¢ Network is programmed by adjusting weights using feedback from examples
- 6. Back to McCulloch-Pitts example... Given that:
- 7. Back to McCulloch-Pitts example... Given that:
- 8. Back to McCulloch-Pitts example... is >= 0 (if the activation function called step function is used)
- 9. Generalization of McCulloch-Pitts Processing Unit
- 12. When to leverage ANN?
- 14. Single-layer NN vs. Multi-layer NN
- 15. ANN Learning
- 16. ANN Learning Using Gradient Descent
- 17. ANN Learning Using Gradient Descent For an excellent step-by-step tutorial on Gradient Descent: https://mccormickml.com/2014/03/04/gradient-descent-derivation/
- 18. Multi-layer NN
- 19. Multi-layer NN
- 20. NN with Two Layers
- 21. NN with Two Layers 3 a4 =
- 22. Backpropagation 1. Computes the error term for the output units using the observed error. 2. From the output layer, repeat: ā¢ propagating the error term back to the previous layer, and ā¢ updating the weights between the two layers until the earliest hidden layer is reached.
- 23. Combining NN layers can create difficult shapes
- 24. Intuition: Multilayered neural networks
- 25. Intuition: Multilayered neural networks
- 26. Intuition: Multilayered neural networks
- 27. Intuition: Multilayered neural networks
- 28. Intuition: Multilayered neural networks
- 29. Intuition: Multilayered neural networks
- 30. Intuition: Multilayered neural networks
- 31. Intuition: Multilayered neural networks
- 32. Intuition: Multilayered neural networks
- 33. Intuition: Multilayered neural networks
- 34. Intuition: Multilayered neural networks
- 35. Intuition: Multilayered neural networks
- 36. Neural network in action: Iris classification
- 37. Neural network in action: Iris classification
- 38. Neural network in action: Iris classification
- 39. Neural network in action: Facial recognition
- 40. Neural network in action: Facial recognition
- 41. Neural network in action: Facial recognition
- 42. Neural network in action: Facial recognition
- 43. Neural network in action: AlphaGo More info about CNN: https://towardsdatascience.com/the-most-intuitive-and-easiest-guide-for-convolutional-neural-network-3607be47480
- 44. Neural network in action: Self-driving car More info about CNN: https://towardsdatascience.com/the-most-intuitive-and-easiest-guide-for-convolutional-neural-network-3607be47480
Editor's Notes
- #2: https://www.pexels.com/photo/galleon-ship-photo-under-the-cloudy-sky-1050656/ References: https://www.youtube.com/watch?v=P2HPcj8lRJE https://imada.sdu.dk/~rolf/Edu/DM534/E18/DM534-marco.pdf bigdata.black
- #3: http://ml.informatik.uni-freiburg.de/former/_media/documents/teaching/ss09/ml/perceptrons.pdf Wikipedia PS: - neuron = sel saraf
- #4: References: https://imada.sdu.dk/~rolf/Edu/DM534/E18/DM534-marco.pdf
- #7: it's not strict either >= or >
- #8: it's not strict either >= or >
- #10: Diasumsikan untuk node i Penjelasan: Input diterima 'neuron' Input diakumulasi dengan input function Melalui fungsi aktivasi, dihasilkan output a_i Fungsi aktivasi $g$ bisa berupa fungsi sigmoid, fungsi step/threshold, dsb Neural network merupakan kumpulan unit atau node (dari unit input hingga unit output) yang terhubung dan membentuk topologi neuron
- #11: Step function = binary Sigmoid function = non-binary (70% activated, 10% activated, etc)
- #12: Rectified Linear Unit https://analyticsindiamag.com/most-common-activation-functions-in-neural-networks-and-rationale-behind-it/
- #13: Input bersifat high-dimensional. Memodelkan relasi yang non-linear dan kompleks. Proses untuk mendapatkan hasil (atau interpretability) tidak penting = black box
- #14: a0 = -1 We assume step function = 0 if x < 0 and 1 otherwise
- #21: Suppose the nodes are Node 3 and Node 4
- #22: Suppose the nodes are Node 3 and Node 4
- #23: https://www.slideshare.net/keepurcalm/backpropagation-in-neural-networks https://www.guru99.com/backpropogation-neural-network.html
- #25: https://www.youtube.com/watch?v=BR9h47Jtqyw
- #26: Non-linear regions
- #27: Combining regions
- #28: Combining regions
- #29: Combining regions
- #30: Combining regions
- #31: Combining regions
- #32: Combining regions
- #33: Combining regions
- #34: Combining regions
- #35: Combining regions with different weights (not best solution though)
- #36: Deep NN
- #40: https://www.nature.com/news/computer-science-the-learning-machines-1.14481
- #41: https://www.nature.com/news/computer-science-the-learning-machines-1.14481
- #42: https://www.nature.com/news/computer-science-the-learning-machines-1.14481
- #43: https://www.nature.com/news/computer-science-the-learning-machines-1.14481
- #45: Rather simplification