Forthcoming Version¶
This release includes the following CI improvements:
Upgrade to Ubuntu Noble (24.04).
Remove Ubuntu Focal (20.04) from the CI.
This release will include a new AML-IP node:
Fiware: node that interacts with FIWARE Context Broker to handle inference data.
Version v0.2.0¶
This release adds new features:
Support Fast CDR v2.2.0.
Implement asynchronous request model in
ModelManagerReceiver
.Add
fastdds.application.id
property to participants and endpoints.Add
fastdds.application.metadata
property to participants and endpoints.Add
MainNode
constructor withdomain
parameter in Python bindings.Add Python bindings for Agent Nodes.
Rename
agent_tool
package toamlip_agent
.
This includes the following Bugfixes:
Update
AsyncComputingNode
andAsyncInferenceNode
to only stop if the current state is running and only run if the current state is stopped.ASAN (Address Sanitizer) fixes.
Fix allowlist namespacing in
AgentNode
.Call change status in
AsyncComputingNode
.
This release includes the following CI improvements:
Migrate CI actions to use eProsima-CI.
Include branch environment variables in CI configuration.
This release add new Documentation features:
Add instructions to build the Docker image.
Add
Agent Tool
section.Add
Enabling Technologies
andInternal Protocols
sections.
This release includes the following Dependencies Update:
Repository |
Old Version |
New Version |
|
---|---|---|---|
Foonathan Memory Vendor |
|||
Fast CDR |
|||
Fast DDS |
|||
Dev Utils |
|||
DDS Pipe |
|||
DDS Router |
Previous Versions¶
Version v0.1.0¶
This is the first release of eProsima AML-IP (Algebraic Machine Learning - Integrating Platform).
This release includes the following User Interface features:
C++ API
Python API
Add implementation of DDS entities.
Add implementation of Multiservice protocol.
Add implementation of Asynchronous Multiservice protocol.
New internal package
amlip_demo_nodes
to include the demos packages.Dockerfile for creating a docker image with AML-IP.
Add Custom RPC communication over DDS.
This release supports the following Deployment Scenarios:
This release includes the following new AML-IP Nodes:
Status: node that listens to other nodes status.
Main: node that sends training data and collects the solution to that data.
Computing: node that waits for training data and retrieves a solution.
Edge: node that sends data and waits for the inferred solution.
Inference: node that waits for data and retrieves an inference.
Model Manager Receiver: node that receives statistical data from models and sends requests to those models.
Model Manager Sender: node that sends statistical data from models, receives requests and sends replies to those requests.
This release includes the following new AML-IP Data Types:
Status: status messages sent by each node with its id, type and current state.
Job: messages that represent training data.
Job Solution: messages that represent the solution for a given set of training data.
Inference: messages that represent a partial data-set.
Inference Solution: messages that represent the inference of a data-set.
Model Statistics: messages that represent statistical data from models.
Model: messages that represent a problem model request.
Model Solution: messages that represent a problem reply with the requested model.
This release includes the following Demos:
This release includes the following Documentation features:
This same Documentation
API Code Documentation
This release includes the following Continuous Integration features:
Continuous Integration deployment in GitHub Actions.
Compile with -Wall flag Clang job.
Add Address Sanitizer check to all tests.
Add Python Liner test to the Python API.
Disable Data Sharing from test.