About this Study Set
This study set covers Artificial Intelligence through
18 practice questions.
A technical assessment covering the history, architectures, and theoretical foundations of AI. Every question includes the correct answer so you can learn as you go — pick any format above to get started.
Questions & Answers
Browse all 18 questions from the
Advanced Concepts in Artificial Intelligence study set below.
Each question shows the correct answer — select a study format above to practice interactively.
1
Which paper introduced the Transformer architecture, which serves as the basis for most modern Large Language Models?
-
A
Attention Is All You Need
-
B
Deep Residual Learning for Image Recognition
-
C
Generative Adversarial Nets
-
D
Mastering the Game of Go with Deep Neural Networks
2
Who is credited with inventing the term 'Artificial Intelligence' at the 1956 Dartmouth Summer Research Project?
-
A
Alan Turing
-
B
John McCarthy
-
C
Marvin Minsky
-
D
Claude Shannon
3
In the context of backpropagation, what is the 'vanishing gradient' problem?
-
A
The loss of data packets during training
-
B
Gradients becoming too small to update weights effectively
-
C
The overfitting of the model to noise
-
D
The hardware overheating during computation
4
Which algorithm is defined as a heuristic search that finds the shortest path by combining the cost to reach a node and the estimated cost to the goal?
-
A
Breadth-First Search
-
B
A* search algorithm
-
C
Dijkstra's algorithm
-
D
Monte Carlo Tree Search
5
What does the 'ReLU' activation function stand for in deep learning?
-
A
Rectified Linear Unit
-
B
Recursive Logical Unit
-
C
Radial Error Linear Update
-
D
Refined Logic Unit
6
Which specific technique is used to prevent overfitting by randomly setting a fraction of input units to 0 at each update during training?
-
A
Bagging
-
B
Dropout
-
C
Boosting
-
D
Pruning
7
What is the primary objective of a 'GAN' (Generative Adversarial Network) architecture?
-
A
To translate languages using recurrent cells
-
B
To have two neural networks compete against each other
-
C
To classify images into thousands of categories
-
D
To map high-dimensional data to lower dimensions
8
In reinforcement learning, what does 'SARSA' stand for?
-
A
State-Action-Reward-State-Action
-
B
Search-Algorithm-Random-State-Analysis
-
C
Statistical-Artificial-Reinforcement-System-Approach
-
D
State-Action-Regression-Sequential-Analysis
9
Which cognitive architecture was developed by John R. Anderson to model human memory and learning?
-
A
SOAR
-
B
ACT-R
-
C
Cyc
-
D
OpenCog
10
What is the 'No Free Lunch Theorem' in machine learning?
-
A
No single model performs best on every possible problem
-
B
Models cannot be trained without labeled data
-
C
Computing power is always limited
-
D
Training time is inversely proportional to accuracy
11
What mathematical operation is the fundamental building block of Convolutional Neural Networks (CNNs)?
-
A
Cross-correlation
-
B
Cross-entropy
-
C
Softmax transformation
-
D
Backpropagation
12
What is the primary limitation of a perceptron that was famously highlighted by Minsky and Papert in 1969?
-
A
Inability to handle non-linear activation functions
-
B
Inability to solve the XOR logical problem
-
C
High computational complexity
-
D
Requirement for exponential training data
13
Which company released the 'AlphaGo' system that defeated Lee Sedol in the game of Go?
-
A
OpenAI
-
B
Google DeepMind
-
C
Microsoft Research
-
D
Facebook AI Research
14
In natural language processing, what are 'word embeddings' like Word2Vec primarily designed to capture?
-
A
Grammatical syntax rules
-
B
Semantic relationships and vector similarity
-
C
The phonetics of spoken language
-
D
The frequency of character occurrence
15
What is the function of the 'Softmax' layer in a neural network?
-
A
To normalize output into a probability distribution
-
B
To compress data into smaller vectors
-
C
To calculate the derivative of the loss function
-
D
To initialize weights randomly
16
Which of these is a widely used benchmark dataset for handwritten digit recognition?
-
A
ImageNet
-
B
MNIST
-
C
CIFAR-10
-
D
COCO
17
What is the 'Transformer' model's 'Self-Attention' mechanism designed to do?
-
A
Weight the importance of different words in a sequence
-
B
Reduce the number of parameters in a network
-
C
Force the network to focus on the start of a sentence
-
D
Increase the depth of a neural network
18
In deep learning, what is a 'Hyperparameter'?
-
A
A weight updated during training
-
B
A parameter whose value is set before the learning process
-
C
The final output of the model
-
D
A node in the hidden layer
