ACtE09 · Chapter 9AI & Neural Networks🏛️ Past Licensure QuestionID: 714ca911
In artificial neural networks, which activation function is standardly used by discrete recurrent Hopfield networks to update binary/bipolar neuron states?
In artificial neural networks, which activation function is standardly used by discrete recurrent Hopfield networks to update binary/bipolar neuron states?
Multiple Choice Options
- A
ReLU (Rectified Linear Unit)
- B
Sigmoid function
- C
Softmax function
- D✓ Correct Option
Sign function (Signum / Threshold activation)
Detailed Technical Solution & Explanation
In a discrete Hopfield network, neurons operate in bipolar states ( or ). The deterministic state update rule calculates the weighted input sum and applies the Signum (Sign / hard threshold) activation function:
where if , and if .
Option Analysis:
•
Option A (ReLU): Used in feedforward deep convolutional networks.
•
Option B (Sigmoid): Used in continuous Hopfield networks and feedforward MLPs.
•
Option C (Softmax): Used for multi-class probability outputs in classifiers.
•
Option D (Sign function): Correct; standard bipolar threshold activation in discrete Hopfield nets.
Reference: Demo2 · Set 2NEC Licensure Syllabus