Unit -1,2
1.Define soft computing.
Soft computing is an approach that uses techniques like fuzzy logic,
neural networks, and genetic algorithms to solve complex problems by
allowing imprecision, uncertainty, and approximation. It mimics human
reasoning to find flexible, adaptive solutions where exact methods fail.
2. Difference between associative memory and hetero associative memory.
Aspect Associative Memory Hetero-Associative Memory
Input and Input and output patterns are Input and output patterns are
Output the same. different.
Recalls complete pattern from Maps input pattern to a related
Function
partial or noisy input. but different output pattern.
"A" → "A" (restore corrupted "A" → "B" (associate "A" with
Example
"A"). related "B").
Also Called Auto-associative memory. Hetero-associative memory.
3. Which are the rules used in Hebb’s law?
Hebb’s Law states that the connection between two neurons
strengthens if they are activated simultaneously.
Rule: Δwij=η⋅xi⋅yj
Where Δwij is the change in weight, η is learning rate, xi is input, and yj is
output.
This rule helps in learning associations in neural networks.
4. What is fuzzy rule?
A fuzzy rule is an IF-THEN statement that relates fuzzy input variables to
fuzzy output variables using linguistic terms.
Example: IF temperature is high THEN fan speed is fast
5. Compare crisp and fuzzy set.
Crisp Set Fuzzy Set
Element either fully belongs Element can partially belong (value
Membership
(1) or not (0). between 0 and 1).
Precision Exact and well-defined boundaries. Vague or imprecise boundaries.
Age ≥ 18 is "adult" (yes or Age 17.5 may be "somewhat
Example
no). adult".
6. Define neural network.
A neural network is a computational model inspired by the structure and
functioning of the human brain. It consists of interconnected nodes (neurons)
organized in layers that process information and learn patterns from data
through training.
Neural networks are widely used in tasks like image recognition, speech
processing, and prediction.
7. List the various types of learning methods used for propagation network.
The various types of learning methods used for propagation networks
(especially in neural networks) include:
1. Supervised Learning
2. Unsupervised Learning
3. Reinforcement Learning
4. Hebbian Learning
5. Error-Correction Learning (e.g., backpropagation)
6. Competitive Learning
8.Draw the basic model of a Adaline network.
9. List the limitations of perceptron.
1. Cannot solve non-linear problems like XOR, as it only works for linearly
separable data.
2. Fixed activation function (step function) limits learning capability and
flexibility
10. estimate the error function and delta rule.
Error Function: The error function measures the difference between the target
output t and the predicted output y typically calculated as: E=1/2(t−y)^2
This function is minimized during training.
Delta Rule: The delta rule updates the weights by adjusting them in the
direction that reduces the error, using: Δwi=η⋅(t−y)⋅xi
Where η is the learning rate, and (t−y)is the error term.
11. List the applications of soft computing.
• Machine Learning: Classification and pattern recognition.
• Control Systems: Fuzzy logic for automation and robotics.
• Medical Diagnosis: Disease detection and analysis.
• Optimization: Solving scheduling and resource allocation problems.
12. What is an expert system?
An expert system is a computer-based application that mimics the
decision-making ability of a human expert. It uses knowledge base (facts and
rules) and inference mechanisms to solve complex problems within a specific
domain, providing advice or solutions.
Example: Medical diagnostic systems.
13. What is fuzzy set?
A fuzzy set is a set where each element has a degree of membership
ranging from 0 to 1, representing the level of truth or belonging to the set.
Unlike classical sets (where an element either belongs or doesn't), fuzzy sets
handle uncertainty and imprecision.
14. What are the types of fuzzy member ship function?
1. Triangular Membership Function
2. Trapezoidal Membership Function
3. Gaussian Membership Function
4. Sigmoidal Membership Function
5. Singleton membership function
6. Generalized bell shaped function
15. Define fuzzy relations.
Fuzzy relations describe the association between elements of two or
more fuzzy sets, where the degree of relation is represented by a value
between 0 and 1
Example:
For two fuzzy sets A and B, a fuzzy relation R can be defined as: R:A×B→[0,1]
16. what is reinforcement learning?
Reinforcement Learning (RL) is a type of machine learning where an
agent learns to make decisions by interacting with an environment. It aims to
maximize cumulative rewards by taking actions that yield positive feedback.
The agent receives rewards or penalties based on its actions and adjusts its
strategy (policy) to achieve the best outcome.
17. what is MADALINE?
MADALINE (Multiple Adaptive Linear Neurons) is an extension of the
Adaline model, consisting of multiple Adaline units organized in layers. It is
trained using the Least Mean Squares (LMS) algorithm and is used for pattern
recognition and classification tasks.
18. What are the applications on BPN?
• Pattern Recognition: Used for recognizing patterns in images, speech,
and handwriting.
• Classification and Prediction: Applied in data classification tasks and
forecasting in areas like stock market prediction.
• Prediction: Applied in time-series forecasting and stock market
prediction.
19. What is content addressable memory?
Content Addressable Memory (CAM) is a type of memory where data is
accessed based on its content rather than its address. The memory is searched
for a specific pattern or value, and the corresponding location is returned.
20. What is local minima and global minima?
Local Minima and Global Minima refer to the lowest points in a function
1. Local Minima: A point where the function value is lower than all its
neighboring points, but not necessarily the lowest overall.
2. Global Minima: The absolute lowest point of the function, where the
value is lower than at all other points in the entire domain
