UGC NET 2024: Artificial Intelligence Questions and Answers with detailed explanation

 UGC NET 2024: Artificial Intelligence Questions and Answers with detailed explanation exams held in June 2024 and Oct 2024.

Q.No: 137.

Answer: 3

Explanation:

A. Natural language processing (NLP) is a machine learning technology that gives computers the ability to interpret, manipulate, and comprehend human language.

NLP stands for Natural Language Processing, a branch of artificial intelligence that uses machine learning to process and understand human language. NLP is a field that combines computer science, computational linguistics, and artificial intelligence.

B. Reinforcement learning (RL) is a machine learning technique that teaches software how to make decisions to achieve the best results. It's a powerful tool for artificial intelligence (AI) systems to learn how to perform tasks in new environments without the help of a supervisor.

 Here are some key characteristics of RL

1. Trial-and-error learning

RL mimics how humans learn by trial and error, where actions that lead to the desired goal are reinforced, and actions that don't are ignored.

2. Reward-and-punishment

RL algorithms use a reward-and-punishment paradigm to process data

3. Delayed gratification

RL algorithms can consider the delayed reward of an action, not just the immediate reward.

4. Goal-oriented

RL agents learn how to achieve their goal by taking the best actions to maximize the reward over a given time period.

5. Episodic or continuous tasks

Tasks can be episodic, with a starting and ending point, or continuous, with no ending point

C. A support vector machine (SVM) is a supervised machine learning algorithm that classifies data by finding the best hyperplane to separate classes in a multidimensional space.

 

SVMs are used for classification, regression, and outlier detection. They are effective in high-dimensional spaces and can handle cases where the number of dimensions is greater than the number of samples. SVMs are also memory efficient because they use a subset of training points in the decision function, called support vectors.

D. Rule-based systems, a foundational technology in artificial intelligence (AI), have long been instrumental in decision-making and problem-solving across various domains. These systems operate on a set of predefined rules and logic to make decisions, perform tasks, or derive conclusions.

1. Expert Systems

Expert systems are designed to emulate the decision-making abilities of human experts. They use a large set of rules to make inferences or recommendations.

Applications include:

• Medical Diagnosis: Systems like MYCIN, an early expert system, assist in diagnosing diseases and recommending treatments based on patient data and medical knowledge.
• Financial Services: Expert systems can assess credit risks, detect fraud, and provide investment advice.

2. Decision Support Systems

These systems aid decision-making processes by providing relevant information and recommendations based on predefined rules. Examples include:

• Customer Support: Automated customer support systems use rules to handle common queries and problems, providing quick and consistent responses.
• Manufacturing: Rule-based systems help in managing production schedules, inventory control, and quality assurance.

3. Control Systems

In control systems, rules govern the operation of machinery and equipment. Examples include:

• Automated Traffic Lights: Rules determine the timing and sequencing of traffic lights based on traffic flow and other factors.
• Home Automation: Smart home systems use rules to control lighting, heating, and security systems based on user preferences and sensor data.

 Q.No: 141 to 145

Read the below passage and answer the questions.

Artificial Neural Networks (ANNs) are computational models inspired by the human brain's neural networks. They consist of inter-connected nodes, or neurons, organized into layers: an input layer, one or more hidden layers, and an output layer. Each connection between neurons has a weight that adjusts as learning progresses, allowing the network to adapt and improve its performance. ANNs are particularly effective in recognizing patterns, making them valuable for tasks such as image and speech recognition, natural language processing, and predictive analytics.

Learning in ANNs typically involves training algorithms like backpropagation, which minimize the error by adjusting the weights. As a subset of machine learning, ANNs have revolutionized the field of Artificial Intelligence by providing solutions to complex problems that traditional algorithms struggle with.

 

Q.No: 141 

Which of the following is/are the application area(s) of ANN?

(A) Natural Language Processing
(B) Image Processing
(C) Pattern Recognition
(D) Speech Recognition

Choose the correct answer from the options given below:

1. (A) and (B) Only
2. (B) and (C) Only
3. (A), (B), and (C) Only
4. (A), (B), (C), and (D)

 Explanation:

From the passage provided earlier:

Artificial Neural Networks (ANNs) are effective in natural language processing (A), image recognition (B), pattern recognition (C), and speech recognition (D). Thus, all the given applications are valid.

Correct answer: (4) (A), (B), (C), and (D).

Q.No: 142

What is the role of the Backpropagation Algorithm?

1. To reduce error
2. To secure network
3. To control speed of data
4. To add different layers

From the earlier passage, backpropagation is described as an algorithm that minimizes the error by adjusting the weights. Its primary role is to reduce error during the training of Artificial Neural Networks (ANNs).

Correct answer: (1) To reduce error.

Q.No: 143

Which of the following layers may be more than one in number?

1. Input layer
2. Hidden layer
3. Output layer
4. Physical layer

From the description of Artificial Neural Networks (ANNs), they consist of an input layer, one or more hidden layers, and an output layer. The hidden layers are the layers that can exist in multiple numbers in a neural network.

Correct answer: (2) Hidden layer.

Q.No: 144

Artificial Neural Networks (ANNs) are inspired by:

1. Quantum mechanics
2. Human brain's neural network
3. Computer hardware architecture
4. Genetic algorithm

From the passage provided earlier, Artificial Neural Networks (ANNs) are computational models inspired by the human brain's neural network.

Correct answer: (2) Human brain's neural network.

Q.No: 145

What is the role of weights in an ANN?

1. To store data
2. To adjust and improve network performance
3. To control the speed
4. To secure the network

Answer: (2) To adjust and improve network performance

.......................................................................................................................................................................

June 2024 

Question 96:
What is a perceptron in ANN?

1. A double-layer auto-associative neural network.
2. A neural network that contains feedback.
3. An auto-associative neural network.
4. A single-layer feed-forward neural network with pre-processing.

Answer:
(4) A single-layer feed-forward neural network with pre-processing

A perceptron is one of the simplest types of artificial neural networks, consisting of a single layer of weights. It is used for linear binary classification tasks.

Question 101:
Backpropagation is a learning technique that adjusts weights in the neural network by propagating weight changes:

1. Forward from source to hidden nodes
2. Backward from sink to source
3. Forward from source to sink
4. Backward from sink to hidden nodes

Answer:
(2) Backward from sink to source

Backpropagation is a supervised learning algorithm where the error is calculated at the output layer (sink) and propagated backward through the network to update the weights from the output layer to the input layer (source).

Question 109:
Arrange the following examples of Artificial Intelligence (AI) in the order of increasing complexity:

A. Spam email detection using rule-based systems.
B. Handwritten digit recognition using shallow neural networks.
C. Image classification using convolutional neural networks.
D. Autonomous driving using reinforcement learning algorithms.

Choose the correct answer from the options given below:

1. A, B, C, D
2. B, A, C, D
3. A, B, D, C
4. B, C, A, D

Answer:
(1) A, B, C, D

Explanation:

• A (Spam email detection): Rule-based systems are the simplest form of AI.
• B (Handwritten digit recognition): Shallow neural networks are a bit more complex.
• C (Image classification): Convolutional neural networks involve deeper architectures, making it more complex.
• D (Autonomous driving): Reinforcement learning is highly complex due to dynamic decision-making in real-time environments.

Question 122:
In Genetic Algorithm's crossover operation, mask is used in:

1. Three parent crossover
2. Two parent crossover
3. Uniform crossover
4. N-point crossover

Answer:
(3) Uniform crossover

In a Uniform crossover, a mask is used to determine which genes (bits) are selected from each parent to create the offspring. The mask specifies the positions where genes are exchanged, providing more randomness compared to other crossover methods.

Here is an explanation of all the options provided in Question 122:

1. Three Parent Crossover

• Explanation: In this method, the genetic material from three parents is combined to produce offspring. Each gene in the offspring is typically computed as an average or a weighted combination of the corresponding genes from the three parents.
• Use of Mask: No mask is used here, as the genes are blended mathematically rather than selected based on specific positions.

2. Two Parent Crossover

• Explanation: This is the most common crossover operation where genes from two parents are combined to produce offspring. Examples include single-point crossover and multi-point crossover.
• Use of Mask: No direct mask is used in basic two-parent crossover methods. Instead, fixed points or random points are used to decide the splitting or mixing of genetic material.

3. Uniform Crossover

• Explanation: In a uniform crossover, a mask determines the selection of genes from each parent. For each position in the offspring, the mask specifies whether the gene comes from Parent 1 or Parent 2.
• Use of Mask: Mask is the key element here. It is typically represented as a binary vector (e.g., 1 for genes from Parent 1 and 0 for genes from Parent 2). This adds more randomness and diversity compared to fixed-point crossovers.
• Example:
  Parent 1: 101010
  Parent 2: 010101
  Mask: 110011
  Offspring: 100011

4. N-Point Crossover

• Explanation: In this method, N points are randomly chosen in the genome, and the genetic material is exchanged between the parents at these points. For example, in a two-point crossover, two points divide the genome into three segments, and alternating segments are exchanged between parents to create offspring.
• Use of Mask: No mask is used. The crossover points are determined randomly, and genes are exchanged based on those points.

Question 128:
The parameter Actual count is used in Genetic Algorithm (GA) for:

1. Crossover
2. Mutation
3. Selecting population
4. Encoding the Genetic Algorithm

Answer:
(3) Selecting population

Explanation:

• In Genetic Algorithms, actual count typically refers to the number of individuals or solutions in the population that are considered at any given stage of the algorithm.
• It is particularly important during the selection phase, where the fittest individuals are chosen based on their fitness scores to form the next generation.
• Other phases like crossover and mutation rely on the selected population for genetic manipulation, but the actual count parameter is primarily associated with selecting the appropriate population size.

Q.No: 129. For the following parents in Figure1 and Figure2

Which is correct?

1. Figure (2) is the two-point crossover of figure (1).
2. Figure (2) is one-point crossover of figure (1).
3. Figure (2) is N-point crossover of figure (1).
4. Figure (2) is uniform crossover of figure (1).

Answer:

(1) Figure (2) is the two-point crossover of figure (1).

Explanation:

• Two-point crossover involves selecting two points on the parents' chromosome strings and swapping the segments between those points.
• In Figure (1), two crossover points are clearly marked, where the first crossover happens at the boundary of the first and second segments, and the second crossover happens at the boundary of the second and third segments.
• The offspring in Figure (2) is generated by swapping the middle segment (indicated by the dashed lines).

Other options are incorrect because:

• One-point crossover swaps segments after a single point, which is not the case here.
• N-point crossover refers to multiple crossover points (more than two), but only two points are shown.
• Uniform crossover relies on a mask to select genes from either parent randomly, which is not evident in this example.

Question 134:

Which of the following structures of knowledge representation are the most purely syntactic?

A. Conceptual dependency
B. Predicate logic
C. Slot-and-filler
D. Nonmonotonic systems

Choose the correct answer from the options given below:

1. A and B Only
2. B and C Only
3. B and D Only
4. C and D Only

Answer:

(3) B and D Only

Explanation:

• Predicate Logic (B): It is purely syntactic because it is based on formal symbols and rules for logical expressions. There is no inherent semantics (meaning); it is only syntactic manipulation of symbols.

• Nonmonotonic Systems (D): These systems are syntactic in nature because they involve changing inference rules or retracting conclusions when new information is introduced. The focus is on the syntax of how rules and knowledge are manipulated.

• Conceptual Dependency (A): This is not purely syntactic, as it focuses on representing meaning and relationships between concepts (semantics).

• Slot-and-Filler (C): This is also not purely syntactic, as it is a semantic knowledge representation model where specific "slots" are filled with "values" that have meaning.

Thus, B and D are the most syntactic approaches in the given options.