Oracle Database 26ai Vector Search

A Vector Database is a specialized database designed to efficiently handle and query high-dimensional Vector data.

Vector databases play a crucial role in Artificial Intelligence (AI) including LLMs (Large Language Models) and Machine Learning (ML) applications, by enabling the efficient storage and retrieval of high-dimensional vector data. They are used in a variety of applications, including:

• Natural language processing

• Computer vision

• Recommendation systems

• Anomaly detection

By leveraging the capabilities of vector databases, AI and ML applications can achieve higher efficiency, accuracy, and scalability, making them indispensable tools in the modern data-driven landscape.

Vector Databases are the next evolution in how we store and retrieve data.

Vector Search enables searching both structured and unstructured data by semantics or meaning, and by values, enabling ultra-sophisticated AI search applications.

Starting with the 23ai release, Oracle Database supports storing of Vector Embeddings along side with other business data. This makes Oracle Database ideally suited for traditional and AI-based storage models.

Native AI vector search capabilities can also help Large Language Models (LLMs) deliver more accurate and contextually relevant results for enterprise use cases using retrieval-augmented generation (RAG) on business data.

This also mandates every traditional Oracle DBA & Developer to master the Vector Database and Vector Search skills to stay up to date & relevant in their Career...

In this course, you will leverage the key capability of Oracle Database 26ai to design and manage Artificial Intelligence (AI) workloads using the new Oracle AI Vector Data type and Vector Search feature.

You will learn how to create tables with vector data type, load data, and the query them based on semantics, rather than keywords. You will learn to perform semantic search on unstructured data by combining it with your relational data in one single system. With hands-on practices, you'll be be able to reinforce the learning of the new AI Vector Search feature and its capabilities.

After completion this course, you should be able to:

• Describe a Vector Database and the use cases for Vector databases like Oracle Database 26ai

• Understand how Vector Databases work

• Distinguish Vector database from the Traditional databases

• Key Advantages of Vector Databases

• Describe AI Vector Search features, benefits, and capabilities

• Understand Oracle AI Vector Search Workflow

• Run basic queries on vectors

• Perform DML and DDL operations on vectors

• Upload Vector Embedding Models into Oracle Database

• Generate Vector Embeddings using AI Models

• Store Embeddings in Oracle Tables

• Create Vector Indexes

• Use SQL Functions for Vector Operations

• Perform Vector Search

After completing this course, you'll be able to equip yourself with future-proof skills in AI-powered data management, making you a valuable asset in the evolving tech landscape

Course Overview

• What you will Learn?

• Audience

• Benefits

• Requirements

Overview of Vector Databases

• What is Vector Database?

• What are Vector Embeddings?

• What is Vector Search?

• Advantages of Vector Search

• How does Vector Database work?

• Difference between Vector Database and Traditional Database

• Role of Vector Database in AI & ML application development

• Examples of Vector Databases

Preparing the Practice Environment

• Download and Import Pre-built VM

• Perform Sanity Checks

• Become Familiar with Practice Environment

Overview of Oracle Vector Search

• Overview of Oracle AI Vector Search

• Why use Oracle AI Vector Search

• Oracle AI Vector Search Workflow

Becoming Familiar with Vector Data and Vector Operations

• Creating Table with Vector Data Type Column

• Inserting Vector Data

• Selecting Vector Data

• Perform DDL, DML Operations on Vector Data

• Prohibited Operations

Generate Vector Embeddings

• About Vector Generation

• Import Pretrained Models in ONNX format

• Access Third-party Models using REST APIs

Store Vector Embeddings

• Create Tables using VECTOR Data Type

• Insert Vectors into tables using INSERT statement

• Load Vector Data using SQL*Loader

• Unload and Load Vectors using Oracle Data Pump

Create Vector Indexes and Hybrid Vector Indexes

• What are Vector Indexes?

• Why Vector Index?

• In-Memory Neighbor Graph Vector Index

• Neighbor Partition Vector Index

• Sizing the VECTOR POOL

• Guidelines for using Vector Indexes

• Hybrid Vector Indexes

• When to use Hybrid Vector Index?

Use SQL Functions for Vector Operations

• Vector Distance Functions

  • VECTOR_DISTANCE

  • L1_DISTANCE

  • L2_DISTANCE

  • COSINE_DISTANCE

  • INNER_PRODUCT

  • HAMMING_DISTANCE

  • JACCARD_DISTANCE

  • VECTOR

  • TO_VECTOR

  • VECTOR_NORM

  • VECTOR_DIMENSION_COUNT

  • VECTOR_DIMS

  • VECTOR_DIMENSION_FORMAT

• Vector Distance Metrics

  • Euclidean and Euclidean Squared Distances

  • Cosine Similarity

  • Dot Product Similarity

  • Manhattan Distance

  • Hamming Distance

  • Jaccard Similarity

Query Data with Similarity and Hybrid Searches

• Perform Exact Similarity Search

• Perform Approximate Similarity Search using Vector Indexes

• Perform Multi-Vector Similarity Search

• Perform Hybrid Search