Title Page
Abstract
Contents
I. Introduction 14
1.1. Research Background 14
1.2. Research Goal 16
1.3. Research Contents 17
II. Related Work 19
2.1. Neuron Overview 19
2.1.1. Previous Research 20
2.1.2. Neurons Structure 23
2.1.3. Neurons Types 28
2.1.4. neurons model 31
2.1.5. Neuronal Action Potential 34
2.2. Interneurons 38
2.3. Synaptic Plasticity 39
2.3.1. The synaptic model of neurons 42
2.3.2. Hebb's theory 45
2.3.3. Learning mechanism of synaptic plasticity 47
2.4. Neural Networks 49
2.4.1. Existing neural networks 54
2.4.2. Biological Neural Network 57
2.5. Gill Withdrawal Reflex 59
2.5.1. Why choose Aplysia as the research object 59
2.5.2. Gill withdrawal reflex 62
2.5.3. The research process of gill reflex 63
III. Thesis research content 70
3.1. Hypothesis 70
3.2. Experiment 72
3.2.1. Kandel'sexperiment 72
3.2.2. Network experiment 76
IV. Experimental Result 78
4.1. Build a neural network model 79
4.2. Adjust the neural network model 83
4.3. The effect of the neural network model when deleting a neuron 93
4.3.1. When x₁¹ node is deleted 94
4.3.2. When x₂¹ node is deleted 97
4.3.3. When x₃¹ node is deleted 100
4.3.4. When x₄¹ node is deleted 103
4.4. Network performance when deleting a node in a single learning form 107
4.4.1. Experimental results when a node is deleted during the habituation process 108
4.4.2. Experimental results when a node is deleted during the sensitization process 112
4.4.3. Experimental results when a node is deleted during the classical conditioning process 118
V. Conclusion 128
References 130
Table 1. Comparison of disadvantages of several existing neural networks 56
Table 2. Three manifestations and the intensity of the gill withdrawal reflex 81
Table 3. Expected simulation results 90
Table 4. Final experimental results 126
Fig.2.1. Camillo Golgi (1843-1926) 21
Fig.2.2. Santiago Ramóny Cajal(1852~1934) 22
Fig.2.3. Components of neuron 24
Fig.2.4. The synapse 26
Fig.2.5. Divergence 27
Fig.2.6. Convergence 28
Fig.2.7. Three types of neuron 29
Fig.2.8. Basic Neuron Types 31
Fig.2.9. M-P neuron model 32
Fig.2.10. Neuron ion channel 35
Fig.2.11. The process of cell membrane action potential generation 36
Fig.2.12. Interneuron 38
Fig.2.13. A neuron A that is postsynaptic to two other neurons 46
Fig.2.14. Gill withdrawal reflex circuit 48
Fig.2.15. Artificial Neural Networks 50
Fig.2.16. (A) Biological Neurons and Information Transmission.... 52
Fig.2.17. biological + Artificial Neural Networks 59
Fig.2.18. Aplysia California 60
Fig.2.19. Micrograph of the Aplysia dorsal ganglion 61
Fig.2.20. Aplysia gill withdrawal reflex and siphon... 62
Fig.2.21. Habituation response 64
Fig.2.22. The enhanced response to a single stimulus lasts only a few hours,... 67
Fig.2.23. classical conditioning 68
Fig.3.1. The relationship between logistic regression algorithm, loss... 71
Fig.3.2. Biological neuron + artificial neural network 72
Fig.3.3. Cellular mechanism of learning and memory in Aplysia 73
Fig.4.1. The basic structure of the model 80
Fig.4.2. Model test results 82
Fig.4.3. Adjusted neural network model 84
Fig.4.4. Experimental results after adjusting the model 86
Fig.4.5. Aplysia gill withdrawal reflex circuit 88
Fig.4.6. Artificial neural network model of gill withdrawal reflex 89
Fig.4.7. Gill withdrawal reflex artificial neural network model experiment simulation effect. 92
Fig.4.8. Network model and gill withdrawal reflex circuit 94
Fig.4.9. Experimental result when x₁¹ node is deleted 95
Fig.4.10. when x₁¹ node is deleted 96
Fig.4.11. Experimental result when x₂¹ node is deleted 98
Fig.4.12. when x₂¹ node is deleted 99
Fig.4.13. Experimental result when x₃¹ node is deleted 101
Fig.4.14. when x₃¹ node is deleted 102
Fig.4.15. Experimental result when x₄¹ node is deleted 104
Fig.4.16. when x₄¹ node is deleted 105
Fig.4.17. Normal results under habituation 108
Fig.4.18. when node 2 is deleted 110
Fig.4.19. (A)when node 4 is deleted 111
Fig.4.19. (B)when node 4 is deleted 112
Fig.4.20. Normal results under sensitization 113
Fig.4.21. (A)when node 2 is deleted 114
Fig.4.22. when node 3 is deleted 116
Fig.4.23. (A)when node 4 is deleted 117
Fig.4.24. Normal results under classical conditioning 119
Fig.4.25. when node 1 is deleted 121
Fig.4.26. when node 2 is deleted 123
Fig.4.27. (A)when node 4 is deleted 124