AI General Terms Flashcards

Improve base knowledge

1
Q

Neural Network

A

a computer system modeled on the human brain and nervous system.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is Grounding in AI?

A

The process of connecting language models to factual, verifiable information and real-world data sources, ensuring outputs are based on accurate, current information rather than training data alone.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

How does Google Cloud Platform implement Grounding?

A

GCP implements grounding through:

Vertex AI’s built-in data connectors
Integration with enterprise data sources
Real-time data validation
Ability to connect to external APIs and databases

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What is Tuning in machine learning?

A

The process of adjusting model parameters during training to optimize performance. Includes both automated parameter adjustment during training and manual hyperparameter optimization.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

How does Fine Tuning differ from regular tuning?

A

The process of taking a pre-trained model and further training it on a specific dataset for a particular task or domain. This allows the model to specialize while retaining its base knowledge.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What does Model Management encompass in GCP?

A

Comprehensive oversight of ML models including:

Version control
Deployment management
Resource allocation
Access control
Model lifecycle management

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What aspects of ML models should be monitored?

A

Key monitoring aspects include:

Model performance metrics
Prediction quality
Resource utilization
Data drift
Model drift
Latency and throughput
Error rates

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What is Prompt Management?

A

The systematic approach to:

Creating and organizing prompts
Versioning prompt templates
Testing prompt effectiveness
Measuring prompt performance
Standardizing prompt patterns across applications

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What should be tracked in model Notes and Status?

A

Critical tracking elements:

Training history
Performance metrics
Deployment status
Known issues
Update history
Dependencies
Production readiness

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What should be included in ML model change tracking?

A

Essential elements to track:

Code changes
Data updates
Parameter modifications
Performance impact
Environmental changes
Deployment status
Rollback points

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is the Softmax function?

A

A mathematical function that converts a vector of numbers into a probability distribution. Commonly used in neural networks’ output layer for multi-class classification, ensuring all probabilities sum to 1.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What is RAG?

A

Retrieval-Augmented Generation

A technique that enhances language model responses by:

Retrieving relevant information from external sources
Incorporating this information into the generation process
Providing current and accurate information
Maintaining source attribution

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What are RNNs and their use cases?

A

Neural networks designed for sequential data processing:

Maintains internal memory state
Processes sequences one element at a time
Suitable for time series, text, and speech
Can handle variable-length inputs

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What are CNNs and their primary applications?

A

Neural networks specialized for processing grid-like data:

Excellent for image processing
Feature detection through convolution operations
Spatial hierarchy learning
Reduced parameter count compared to fully connected networks

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What is a Vector in ML context?

A

A mathematical representation of data points:

Ordered array of numbers
Represents features in multi-dimensional space
Used for embeddings in ML models
Enables similarity comparisons

Like words from a page represented in numbers
Like a set of numbers representing a picture
0’s adn 1’s

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is Few Shot Learning?

A

The ability of a model to learn from very few examples:

Requires minimal training examples
Leverages transfer learning
Uses meta-learning techniques
Particularly useful for rare cases or new categories

17
Q

hat is Sentiment Analysis?

A

ML technique to determine emotional tone in text:

Classifies text as positive, negative, or neutral
Uses NLP techniques
Can detect emotional nuances
Common in customer feedback analysis

18
Q

What is a Tensor?

A

Multi-dimensional array used in ML:

Generalizes vectors and matrices
Basic data structure in deep learning
Represents complex data relationships
Core component in frameworks like TensorFlow

19
Q

What is Apache Beam?

A

Unified programming model for:

Batch and streaming data processing
Portable across execution engines
Pipeline-based processing
Supported natively in GCP Dataflow

20
Q

What is Apache Airflow?

A

Platform to programmatically author, schedule, and monitor workflows:

Creates DAGs of tasks
Manages task dependencies
Handles retry logic
Monitors execution

21
Q

What is a DAG in data processing?

A

A workflow representation where:

Tasks are nodes
Dependencies are directed edges
No cycles allowed
Defines processing order

22
Q

What is Data Fabric?

A

Architecture that:

Integrates data sources
Provides unified data management
Enables consistent security
Supports data governance

23
Q

What is a Data Lake?

A

Storage repository that:

Holds raw data in native format
Supports structured and unstructured data
Enables big data analytics
Scales horizontally

24
Q

What is a Lake House architecture?

A

Hybrid architecture combining:

Data lake flexibility
Data warehouse performance
ACID transactions
Schema enforcement when needed

25
Q

What are Hyperparameters?

A

Configuration settings used to control the learning process:

Set before training begins
Not learned from data
Examples: learning rate, batch size
Tuned through experimentation