AI, ML & NLP

Artificial Intelligence (AI)

Definition: AI is the simulation of human intelligence by machines.

  • Types of AI:
    1. Narrow AI: Specialized tasks (e.g., Siri, Google Translate).
    2. General AI: Broad human-like capabilities (theoretical).
    3. Superintelligent AI: Surpasses human intelligence (future possibility).

Key Concepts:

  • Machine Learning (ML): AI that learns from data.
  • Reasoning and Problem Solving: E.g., Chess-playing AI.
  • Perception: Image recognition.
  • Natural Language Processing (NLP): Language understanding and generation.

Machine Learning (ML)

Definition: Algorithms learn patterns from data without explicit programming.

Types of ML:

  1. Supervised Learning:
    • Data with labels.
    • Examples: Spam detection, predicting house prices.
    • Algorithms: Linear Regression, Decision Trees, SVM.
  2. Unsupervised Learning:
    • Data without labels.
    • Examples: Clustering customers, dimensionality reduction.
    • Algorithms: K-means, PCA.
  3. Reinforcement Learning:
    • Learning by trial and error (rewards/punishments).
    • Example: Game-playing bots (AlphaGo).

Key Concepts:

  • Overfitting: Model too focused on training data; poor generalization.
  • Underfitting: Model too simple; misses key patterns.
  • Hyperparameters: Parameters tuned for better performance.

Important Algorithms:

  • Linear Regression: Predicts a continuous value.
  • Logistic Regression: Predicts probabilities (classification).
  • Neural Networks: Layers of nodes mimic the brain (deep learning).

Natural Language Processing (NLP)

Definition: Field of AI focused on interaction between computers and human language.

Key Tasks:

  1. Text Classification: Spam detection, sentiment analysis.
  2. Tokenization: Breaking text into words/tokens.
  3. POS Tagging: Identifying parts of speech (noun, verb).
  4. Named Entity Recognition (NER): Detecting names, dates, etc.
  5. Machine Translation: Translating text (Google Translate).
  6. Speech Recognition: Converting speech to text.
  7. Text Generation: Chatbots, story generation.

Techniques:

  • Bag of Words (BoW): Simplistic text representation.
  • TF-IDF: Scores word importance in a document.
  • Word Embeddings: Vector representation of words (e.g., Word2Vec, GloVe).
  • Transformer Models: Advanced architectures (e.g., BERT, GPT).

Applications:

  • Chatbots (e.g., Alexa).
  • Sentiment Analysis (e.g., analyzing tweets).
  • Text Summarization.

Key Terms to Remember

  • Deep Learning: Advanced ML using neural networks.
  • Gradient Descent: Optimization technique in training ML models.
  • AI Ethics: Bias, accountability, data privacy issues in AI.
  • Turing Test: Test for machine’s human-like intelligence.
  • Confusion Matrix: Tool to evaluate classification models.
  • Precision & Recall: Metrics for evaluating ML models.

Mnemonics for Quick Recall

  • Types of AI: Narrow, General, Super (NGS).
  • ML Types: Supervised, Unsupervised, Reinforcement (SUR: See Us Run).
  • NLP Key Tasks: Classify, Tokenize, POS, Recognize, Translate, Speak, Generate (CTPRTSG: Cats Take Pretty Roses To Sweet Gardens).

MCQ

Artificial Intelligence (AI)

What is Artificial Intelligence (AI)?
a) Making a computer intelligent
b) Programming with high-level languages
c) Automating manual tasks
d) Hardware improvement

a

Which of the following is NOT a type of AI?
a) Narrow AI
b) General AI
c) Superintelligent AI
d) Self-supervised AI

d

The test to determine whether a machine exhibits intelligent behavior equivalent to a human is called:
a) Turing Test
b) Alan Test
c) AI Benchmark Test
d) Neural Validation Test

a

Which programming language is most widely used in AI?
a) Java
b) Python
c) C++
d) Perl

b

In AI, what is the primary role of heuristic functions?
a) To reduce computation time
b) To ensure guaranteed solutions
c) To handle large datasets
d) To optimize hardware utilization

a

Machine Learning (ML)

Machine Learning is a subfield of:
a) AI
b) Data Science
c) Statistics
d) All of the above

Answer: d

Which ML approach uses labeled data for training?
a) Supervised Learning
b) Unsupervised Learning
c) Reinforcement Learning
d) Clustering

a

Which algorithm is used for classification problems?
a) Linear Regression
b) Logistic Regression
c) K-means Clustering
d) PCA

b

Overfitting in machine learning can be reduced by:
a) Using more data
b) Early stopping
c) Regularization techniques
d) All of the above

d

What is the main objective of reinforcement learning?
a) To cluster data
b) To predict continuous outcomes
c) To maximize a reward signal
d) To reduce dimensionality

c

Which evaluation metric is most suitable for imbalanced datasets?
a) Accuracy
b) Precision and Recall
c) Mean Absolute Error
d) F1 Score

d

Which of the following is a common algorithm in unsupervised learning?
a) K-means
b) Decision Tree
c) Neural Networks
d) Naive Bayes

a

Deep learning is based on:
a) Decision Trees
b) Neural Networks
c) Clustering Algorithms
d) SVM

b

Natural Language Processing (NLP)

NLP focuses on the interaction between:
a) Computers and physical machines
b) Computers and human language
c) Computers and statistical algorithms
d) Humans and robots

b

Which of these is a core task of NLP?
a) Sentiment Analysis
b) Tokenization
c) Part-of-Speech (POS) Tagging
d) All of the above

d

What does TF-IDF stand for in NLP?
a) Term Frequency-Inverse Document Frequency
b) Token Frequency-Inverse Density Factor
c) Text Frequency-Indexed Data Format
d) None of the above

a

What is the main advantage of using Word Embeddings in NLP?
a) They provide context to words
b) They increase data size
c) They simplify data pre-processing
d) None of the above

a

Which model is most commonly used for machine translation tasks?
a) RNN
b) CNN
c) Transformer
d) KNN

c

BERT is an example of:
a) A Word Embedding Technique
b) A Transformer-based model
c) A Statistical NLP Model
d) A Clustering Algorithm

b

The output of a Named Entity Recognition (NER) system is:
a) Parts of Speech Tags
b) Named entities like locations, names, and dates
c) Sentiment scores
d) Grammatical errors

b

What does Gradient Descent optimize in ML?
a) Hyperparameters
b) Loss function
c) Training time
d) Memory usage

b

What is the function of a confusion matrix in classification?
a) To measure prediction errors
b) To identify input-output mappings
c) To reduce dimensionality
d) To calculate training speed

a

Which of these is a challenge in AI ethics?
a) Bias in data
b) Lack of transparency in algorithms
c) Privacy concerns
d) All of the above

d

What is the primary purpose of PCA in machine learning?
a) Feature extraction and dimensionality reduction
b) Classification of data
c) Increasing model accuracy
d) Parameter tuning

a

What does the term “hyperparameter tuning” refer to?
a) Adjusting internal model weights
b) Selecting the best settings for a model
c) Measuring model accuracy
d) Optimizing training datasets

b

What is the primary difference between a neural network and a deep neural network?
a) A neural network has fewer layers, while a deep neural network has more layers.
b) A neural network uses more data, while a deep neural network uses less data.
c) A neural network is used for classification, while a deep neural network is used for regression.
d) A neural network is used for regression, while a deep neural network is used for classification.

a

Which of the following is an example of a deep learning architecture?
a) Convolutional Neural Network (CNN)
b) Recurrent Neural Network (RNN)
c) Long Short-Term Memory (LSTM)
d) All of the above

d

What is the term for the process of training a deep neural network using a large dataset?
a) Batch Gradient Descent
b) Stochastic Gradient Descent
c) Mini-Batch Gradient Descent
d) Transfer Learning

c

What is the primary goal of Natural Language Processing (NLP)?
a) To develop chatbots
b) To improve language translation
c) To enable computers to understand human language
d) To develop speech recognition systems

c

What is the term for the process of converting text data into numerical data? a) Tokenization
b) Stemming
c) Lemmatization
d) Vectorization

d

What is the primary goal of Robotics?
a) To develop autonomous vehicles
b) To improve manufacturing processes
c) To enable computers to interact with the physical world
d) To develop humanoid robots

c

Which of the following is an example of a computer vision task?
a) Image Classification
b) Object Detection
c) Image Segmentation
d) All of the above

d

What is the term for the process of using cameras and sensors to enable robots to perceive their environment?
a) Computer Vision
b) Machine Learning
c) Sensorimotor Control
d) Robotics

a

2019 IBPS IT Officer Exam

  1. What is the primary goal of Artificial Intelligence?
    • Answer: To create intelligent machines.
  2. What is the difference between Machine Learning and Deep Learning?
    • Answer: Machine Learning is a broader concept that includes various algorithms that allow computers to learn from data. Deep Learning is a subset of Machine Learning that uses neural networks with many layers (deep networks) to analyze various factors of data.
  3. What is the term for a machine’s ability to learn from data without being explicitly programmed?
    • Answer: Machine Learning.

2018 IBPS IT Officer Exam

  1. What is the primary application of Natural Language Processing (NLP)?
    • Answer: To enable computers to understand, interpret, and respond to human language in a valuable way.
  2. What is the difference between Supervised and Unsupervised Learning?
    • Answer: Supervised Learning uses labeled data to train models, while Unsupervised Learning uses unlabeled data to find patterns or groupings.
  3. What is the term for a type of neural network that is used for image recognition?
    • Answer: Convolutional Neural Network (CNN).

2017 IBPS IT Officer Exam

  1. What is the primary goal of Robotics?
    • Answer: To enable computers and machines to interact with the physical world and perform tasks autonomously.
  2. What is the term for a type of machine learning algorithm that is used for classification?
    • Answer: Classification algorithms (e.g., Decision Trees, Support Vector Machines, etc.).
  3. What is the difference between a Neural Network and a Deep Neural Network?
    • Answer: A Neural Network typically has one or two layers, while a Deep Neural Network has multiple layers (deep architecture) that allow it to learn complex patterns.

2016 IBPS IT Officer Exam

Answer: A Decision Tree is a single tree structure used for making decisions based on feature values, while a Random Forest is an ensemble of multiple decision trees that improves accuracy and reduces overfitting.

What is the primary application of Computer Vision?

Answer: To enable machines to interpret and make decisions based on visual data from the world, such as images and videos.

What is the term for a type of machine learning algorithm that is used for regression?

Answer: Regression algorithms (e.g., Linear Regression, Polynomial Regression, etc.).

What is the difference between a Decision Tree and a Random Forest?

Answer: A Decision Tree is a single tree structure used for making decisions based on feature values, while a Random Forest is an ensemble of multiple decision trees that improves accuracy and reduces overfitting.