5. Artificial Intelligence
AI is the simulation of human intelligence
processes by machines. These
processes include, reasoning,
remembering, learning and self-
correction.
@Sherrrylst
42. How to use Cognitive Services?
@Sherrrylst
1. Create service in Azure Portal
(https://portal.azure.com/)
43. How to use Cognitive Services?
@Sherrrylst
1. Create service in Azure Portal
(https://portal.azure.com/)
2. Customize and train model in the
portal (Optional)
44. How to use Cognitive Services?
@Sherrrylst
1. Create service in Azure Portal
(https://portal.azure.com/)
2. Customize and train model in the
portal (Optional)
3. Call the REST API from your app
45. How to use Cognitive Services?
@Sherrrylst
1. Create service in Azure Portal
(https://portal.azure.com/)
2. Customize and train model in the
portal (Optional)
3. Call the REST API from your app
4. Parse the JSON response
59. Azure Cognitive Services re-cap
@Sherrrylst
• Easy to use (REST API)
• Easy to provision
• Provide pre-trained models
• Possibility to use custom data
• Free training resources
• Detailed documentation
• Create account for free
62. Resources
• Microsoft Azure Cognitive Services: The Big Picture
• The Mojifier (MS Learn)
• The Mojifier (Github)
• Azure Cognitive Services API — I need your clothes, boots and your motorcycle
• Add conversational intelligence to your apps by using LUIS (MS Learn)
• Discover sentiment in text with the Text Analytics API (MS Learn)
• Create Intelligent Bots with the Azure Bot Service (MS Learn)
• Getting started with Azure Cognitive Services
63. AI Services
Cognitive Services Bot Service Azure ML, Databricks, HDInsight
Pre-build AI Conversational AI Custom AI
Azure Infrastructure
CPU, FPGA, GPU
Cosmos
DB
SQL DB Data Lake IOT EdgeDSVMSparkBatch AISQL DW
64. Deep learning Frameworks
Cognitive Toolkits Tensorflow Azure ML, Databricks, HDInsight
Pre-build AI Conversational AI Custom AI
Deep learning Frameworks
CPU, FPGA, GPU
Cosmos
DB
SQL DB Data Lake IOT EdgeDSVMSparkBatch AISQL DW
Notes de l'éditeur
Devices that can remember, learn, understand and recognize things
ML is all about the ability to learn. Applications that can learn without hardcoding different scenarios.
ML is used in many applications to detect the patterns, Is this a cat or a dog.
In order to detect these patterns, you need to use different techniques.
ANN: Mimics the way that human brain works
DL: Learn from many layers of analysis where each layer has the input from the previous layer
AI is the overall concept to make computers intelligent
You use Machine learning to analyze the data
You use Machine learning to analyze the data
This data contains patterns. How a cat looks like or How a dog looks like
You analyze this data with Machine learning algorithm
To find patterns
With those we create a Model, which is the outcome of this process
Model is the thing to recognizes the patterns
This process is also called training a model
Now application can enter data to see if it can recognize a pattern.
Preparing a set of data with diversity and covers the edge cases
Creating the algorithms and choosing the techniques can be challenging. Also testing the outcome and making sure we get the right result is also super challenging.
This is not the most difficult way, but still it’s challenging to find a secure way with having the performance in mind
Cognitive Services are RESTful APIs that exposes ML models to the outside world.
Azure makes sure that the Cognitive Services are available, secured and perform well. We just need to communicate with the API
You can use either available data or upload your own data
You can use either available data or upload your own data
Computer vision -> Analyze the data and return information like 4 storm troopers standing and sky is blue
Video Indexer -> Analyze the video and extract the text and recognizes things that are in the video
Face -> Detect faces and extract information about the face
Content Moderator -> Video, images or text contains unacceptable content
Custom Vision -> Upload different pictures of Princess Lea for training, so it can recognizes princess Lea
Computer vision -> Analyze the data and return information like 4 storm troopers standing and sky is blue
Video Indexer -> Analyze the video and extract the text and recognizes things that are in the video
Face -> Detect faces and extract information about the face
Content Moderator -> Video, images or text contains unacceptable content
Custom Vision -> Upload different pictures of Princess Lea for training, so it can recognizes princess Lea
Computer vision -> Analyze the data and return information like 4 storm troopers standing and sky is blue
Video Indexer -> Analyze the video and extract the text and recognizes things that are in the video
Face -> Detect faces and extract information about the face
Content Moderator -> Video, images or text contains unacceptable content
Custom Vision -> Upload different pictures of Princess Lea for training, so it can recognizes princess Lea
Computer vision -> Analyze the data and return information like 4 storm troopers standing and sky is blue
Video Indexer -> Analyze the video and extract the text and recognizes things that are in the video
Face -> Detect faces and extract information about the face
Content Moderator -> Video, images or text contains unacceptable content
Custom Vision -> Upload different pictures of Princess Lea for training, so it can recognizes princess Lea
Computer vision -> Analyze the data and return information like 4 storm troopers standing and sky is blue
Video Indexer -> Analyze the video and extract the text and recognizes things that are in the video
Face -> Detect faces and extract information about the face
Content Moderator -> Video, images or text contains unacceptable content
Custom Vision -> Upload different pictures of Princess Lea for training, so it can recognizes princess Lea
Speaker recognition -> Identify people based on their speech
Speech Translation -> Listen and translate to text
Speaker recognition -> Identify people based on their speech
Speech Translation -> Listen and translate to text
Language Understanding -> You feeding it with the command and train it to what it means and after that it understands
Text Analytics: Analyze a text in order to get the sentiment of a text (Positive/Negative), detect the language, extract the key phrase from a piece of text and retrieve the topics
Language Understanding -> You feeding it with the command and train it to what it means and after that it understands
Text Analytics: Analyze a text in order to get the sentiment of a text (Positive/Negative), detect the language, extract the key phrase from a piece of text and retrieve the topics
Turns text to questions and answers. It uses in bots.
Bing Custom Search -> Corporate search engine
Bing Entity Search -> Detects pictures or persons to enrich the result
You can even create a guest account to try it without providing Credit card info and no data will be saved after trial is over (7 days)