AI Practitioner MEMORIZE

Commit to memory frequently missed definitions/questions

Regularization parameter

Optimization technique to reduce overfitting; constrain the model

Tokens

Used in natural language processing to represent words or subwords

Embeddings

A representation of items in a continuous vector space, capturing semantic similarities and relationships between them.

Vectors

Represent words or phrases in a continuous vector space, capturing semantic meanings and relationships between them.

Guardrails for Bedrock

Outlines safe and ethical AI usage practices to mitigate risks and ensure responsible development.

Agents for Bedrock

Uses the reasoning of foundation models (FMs), APIs, and data to break down user requests, gathers relevant information, and efficiently completes tasks—freeing teams to focus on high-value work

Watermark detection for Bedrock

Amazon Bedrock provides a watermark detection feature for images generated by the Amazon Titan Image Generator. This feature helps users identify AI-generated content and mitigate the spread of misinformation

Knowledge Base for Bedrock

A comprehensive repository that consolidates relevant information, best practices, and resources to support AI practitioners using Amazon Bedrock.

Good prompting technique

Instructions, context, input data, output indicator

Asynchronous inference

Processing smaller payloads without requiring real-time responses by queuing requests and handling them in the background

Batch inference

Processing multiple requests simultaneously to improve efficiency and reduce latency in AI model predictions.

Real-time inference

Low latency, immediate

Serverless inference

For unpredictable, sporadic workloads

Mean Absolute Error (MAE)

measures the average magnitude of errors in a set of predictions without considering their direction. MAE is typically used in regression tasks to quantify the accuracy of a continuous variable's predictions, not for classification tasks where the outputs are categorical rather than continuous

Correlation matrix

measures the statistical correlation between different variables or features in a dataset, typically used to understand the relationships between continuous variables. A correlation matrix is not designed to evaluate the performance of a classification model, as it does not provide any insight into the accuracy or errors of categorical predictions.

Root Mean Squared Error (RMSE)

metric commonly used to measure the average error in regression models by calculating the square root of the average squared differences between predicted and actual values. However, RMSE is not suitable for classification tasks, as it is designed to measure continuous outcomes, not discrete class predictions

Confusion matrix

Tool specifically designed to evaluate the performance of classification models by displaying the number of true positives, true negatives, false positives, and false negatives. This matrix provides a detailed breakdown of the model's performance across all classes, making it the most suitable choice for evaluating a classification model's accuracy and identifying potential areas for improvement. It provides a comprehensive overview of the model's performance by detailing how many instances were correctly or incorrectly classified in each category. This enables the company to understand where the model is performing well and where it may need adjustments, such as improving the classification of specific material types.

Transformer model

Transformer models use a self-attention mechanism and implement contextual embeddings

Transformer models are a type of neural network architecture designed to handle sequential data, such as language, in an efficient and scalable way. They rely on a mechanism called self-attention to process input data, allowing them to understand and generate language effectively. Self-attention allows the model to weigh the importance of different words in a sentence when encoding a particular word. This helps the model capture relationships and dependencies between words, regardless of their position in the sequence.

Transformer models use self-attention to weigh the importance of different words in a sentence, allowing them to capture complex dependencies. Positional encodings provide information about word order, and the encoder-decoder architecture enables effective processing and generation of sequences. This makes transformers highly effective for tasks like language translation, text generation, and more.

Transfer learning

It allows a model to utilize the knowledge learned from one task or dataset to improve its performance on a new, but related task. Can help optimize performance by adapting insights from the latest data generated by other models. This approach reduces the need for extensive data and computational resources while ensuring that the models benefit from shared knowledge across related domains or tasks.

AWS Trusted Advisor

Provides guidance to help you provision your resources following AWS best practices. It helps optimize your AWS environment in areas such as cost savings, performance, security, and fault tolerance, making it an essential tool for governance in AI systems.

AWS Config

service for assessing, auditing, and evaluating the configurations of your AWS resources. It helps with continuous monitoring and compliance but does not provide the broad optimization and guidance offered by AWS Trusted Advisor.

AWS Audit Manager

helps you continuously audit your AWS usage to assess risk and compliance with regulations and industry standards. It focuses on compliance reporting rather than providing optimization and guidance.

AWS CloudTrail

records AWS API calls and delivers log files for auditing purposes. While it is essential for tracking user activity and compliance, it does not offer the optimization and best practice recommendations provided by AWS Trusted Advisor.

Provisioned throughput

For testing and deploy customized models for Amazon Bedrock (via fine-tuning or continued pre-training), it is mandatory to use Provisioned Throughput.

Epochs

one complete pass of the entire training dataset through the learning algorithm. Essentially, it's the process of feeding the model every single piece of training data once. The number of epochs is a hyperparameter that determines how many times this process is repeated. 

Shapley values

provide a local explanation by quantifying the contribution of each feature to the prediction for a specific instance

PDP (Partial Dependence Plots)

PDP provides a global explanation by showing the marginal effect of a feature on the model’s predictions across the dataset

Sagemaker Ground Truth

Offers the most comprehensive set of human-in-the-loop capabilities, allowing you to harness the power of human feedback across the ML lifecycle to improve the accuracy and relevancy of models. You can complete a variety of human-in-the-loop tasks from data generation and annotation to model review, customization, and evaluation, either through a self-service or an AWS-managed offering.

Sagemaker Clarify

Helps identify potential bias during data preparation without writing code. You specify input features, such as gender or age, and it runs an analysis job to detect potential bias in those features.

SageMaker Model Dashboard

Centralized portal, accessible from the SageMaker console, where you can view, search, and explore all of the models in your account. You can track which models are deployed for inference and if they are used in batch transform jobs or hosted on endpoints. If you set up monitors with Amazon SageMaker Model Monitor, you can also track the performance of your models as they make real-time predictions on live data.

Model invocation logging

collect invocation logs, model input data, and model output data for all invocations in your AWS account used in Amazon Bedrock; allows the company to capture detailed logs of all model requests and responses, including input data, output predictions, and any errors that occur during model execution. This method provides comprehensive monitoring capabilities, enabling the company to effectively track, audit, and troubleshoot model performance and usage.

Amazon Kendra

highly accurate and easy-to-use enterprise search service that’s powered by machine learning (ML). It allows developers to add search capabilities to their applications so their end users can discover information stored within the vast amount of content spread across their company. This includes data from manuals, research reports, FAQs, human resources (HR) documentation, and customer service guides, which may be found across various systems such as Amazon Simple Storage Servic

Amazon Textract

machine learning (ML) service that automatically extracts text, handwriting, layout elements, and data from scanned documents. It goes beyond simple optical character recognition (OCR) to identify, understand, and extract specific data from documents; not a search service.

Amazon SageMaker Data Wrangler

reduces the time it takes to aggregate and prepare tabular and image data for ML from weeks to minutes; simplify the process of data preparation and feature engineering, and complete each step of the data preparation workflow (including data selection, cleansing, exploration, visualization, and processing at scale) from a single visual interface.

Variational autoencoders (VAE)

use two neural networks—the encoder and the decoder. The encoder neural network maps the input data to a mean and variance for each dimension of the latent space. It generates a random sample from a Gaussian (normal) distribution. This sample is a point in the latent space and represents a compressed, simplified version of the input data. The decoder neural network takes this sampled point from the latent space and reconstructs it back into data that resembles the original input.