Technical Documentation
Familiarize yourself with Lumina’s products by reviewing our technical documentation.
Lumina’s Random Contrast Learning
1. Introduction
Access the Lumina OpenAPI Specification here: https://rclapi.lumina247.io
Lumina’s Random Contrast Learning API is available to the general public, the documentation in this section will guide you through:
- Registration
- API Access
- Uploading Training Data
- Starting a Training Session
- Inference from Trained Models
- RCL Showcase
We can’t wait to see what you build!
2. Registration
All Lumina services require an account. To create a new account or sign in to an exisitng account, please use the Lumina Support Portal: INSERT NEW LINK HERE.
For existing accounts, provide your email and password and click ‘Sign In.’
For new users, click ‘Sign Up’ and fill out the form.
3. RCL API Access
To access Lumina’s RCL APIs, navigate to our ‘Contact Us’ page and select ‘API Access’.
Retrive your authorization token for initial authorization, and use the refresh token for long running processes.
RCL API
The base URL for the RCL API is: https://rclapi.lumina247.io
For your convenience, we have exposed the RCL API Open API document.
In order to use the Open API Document, you will need to provide your authorization token.
Authorization
Click the ‘Authorize’ button on the Open API Document and provide your token in the format bearer token.
Click ‘Authorize’
You now have access to use the Open API Document to explore and use various endpoints.
4. Uploading Training Data
The following endpoints should be called in order:
- Provide a file name. This endpoint will return a document ID.
- Provide a file part to upload, the document ID for the file part, and indicate if this is the last part of a file upload.
- After you have provided all parts to your upload, complete the upload.
5. Start Training Session
The following steps outline how to start your training session:
- Provide your training session key.
- Vector size: The vector size to be used when training the model.
- Enable translation or sensor services.
- If training sensor data, such as IOT sensor inputs and outputs, set ‘trainSensorServices’ to ‘true’.
- If training translation services, set ‘trainTranslationServices’ to ‘true’.
6. Inference from Trained Models
Hosting your Models
In order to use your trained models for inference, you will first need to host your model. The following endpoints have been made available for you:
- Start – Launches or strts an inference host.
- Stop – Stops the inference host.
- Reboot – Reboots the inference host.
- Terminate – Terminates the inference host. Please be careful when terminating an inference host. When terminating an inference host, it is not possible to start a terminated host without contactin the Lumina Support Team.
- Status – Returns the status of an inference host.
Inferring from your Models
When first launching an inference host, you may experience a delay while the host is prepared. This is true even in the event that your host status returns a running status. Typically inference is ready within 20 minutes of initial hosting.
Use the follwing endpoint to infer results from your trained models.
Provide:
- Training Session Key
- Priority Type
- Index
- Accuracy
- Specific
- Inference Detail Type
- Search
- PredictLine – Predicting a line of data.
- PredictNext – Predicting next data.
- TranslateLine – Translation models.
- HotWord – Sensor data.
7. Showcase
The Showcase is a place for you to share and take part in the Random Contrast Learning community. Here, you will be able to publish your work to the braoder community and use the derivative works others have shared with the community.
Publishing to the Showcase
Congratulations! You have build RCL-powered models and you’d like to share them with the world. In a few ease steps, you can share your work with the RCL community.
Publish Showcase Request:
Publish Showcase Response:
Revoking Work from the Showcase
Build something better, or ready to retire a showcase item?
Revoke Showcase Request:
Revoke Showcase Response:
Searching Showcase Items
Looking to use a community model? Start here.
Search Showcase Request:
Search Showcase Response:
Showcase Inference
Found a model on the showcase you would like to try out? Inference is the means with which you will communicate with the models you have discovered.
Inference Request:
Inference Response:
Random Contrast Learning Release Notes
Release Notes - 2.4.16
Release 2.4.16 is a major update to RCL Services.
Breaking changes in this release:
- New training sessions are no longer automatically hosted after training. RCL users must choose when to launch their inference host.
- Existing training sessions are no longer automatically hosted. RCL users must choose to launch existing models for inference.
Release Details:
- Distributed Inference Hosting (NEW)
- Users are now able to independetnly host their own inference services for trained models.
- Training
- Performance and stability improvements.
- Inference
- Performance and stability improvements. The Lumina team has improved latency of inference requests significantly in this release.
Release Notes - 2.3.9
Stability improvements
- Introduced the following positive status messages:
- Created – TrainingSessionCreated event.
- DataReceived – New training data uploaded event, occurs when a user uploads new data for training.
- TrainingRequested – Model training requested event, occurs when a user requests a training session.
- TrainingStarted – Model training started event, occurs when a training server becomes available and begins training.
- DataLakeCreationStarted – Datalake creation event, occurs prior to model training.
- DataLakeIndexing – Indexing event triggered, occurs prior to model training.
- DataLakeCreationCompleted – Datalake creation completed successfully event, occurs prior to model training.
- ModelCreationStarted – Model training started event, occurs when training has started.
- ModelCreationCompleted – Model training completed event, occurs when training has completed successfully.
- ModelSaveCompleted – Model has been saved for hosting event, occurs when the model was successfully persisted to storage.
- TrainingCompleted – Training completed event, occurs when the entire training session has completed successfully.
- Introduced the following negative status messages:
- TrainingFailed – Training failed event, occurs when a training session has failed.
- DataLakeCreationFailed – Datalake creation step failed event, occurs when a datalake fails prior to model training.
- ModelCreationFailed – Model training step failed event, occurs when RCL is unable to train on the data provided.
- ModelSaveFailed – Persist model step failed event, occurs when RCL is unable to persist the model for inference hosting.
Known issues
This release does not resolve saturated / unavailable training hosts, that work is still in development, and we anticipate release 2.3.10 or 2.3.11 this to be fully resolved. This means it’s possible to request training and no hosts are available to honor the request. In the event this occurs, you can simply look for the trainingstarted event , if the event has failed to trigger within 10 minutes of request, it’s safe to assume you should request training again.
S4 (See Something Say Something) Mobile Application
1. Introduction
Lumina’s S4 (See Something Say Something) Mobile Application is available for download from the Google Play Store (https://play.google.com/store/apps) for Android devices and at Apple’s App Store (https://www.apple.com/app-store/) for iOS devices.
S4 is a crowdsourced mobile application that helps keep our communities – and each other – safe. See Something Say Something began with a campaign by the New York Metropolitan Transit Authority in 2003, later adopted by the Department of Homeland Security, to inspire, empower, and educate the public to report suspicious activity.
S4 is simple to use and allows users to report suspicious behavior confidentially and in real time.
2. Sign In Options
S4 allows any user to register and create a free account or you may choose to use the app without an account.
If you want to register an account, launch the app and select the top left navigation menu. Then choose Sign In and from the sign in page, select the Sign Up option. When registering a new account, you will be asked to provide an email address, name, phone number and a password.
3. Terms
Acceptance of Terms
When launching the S4 app for the first time, you will be required to acceptance our Terms of Service.
Review the document and select accept if you agree to the Terms.
4. App Navigation
Navigation Menu
Navigation options are located in the top left menu of the S4 app. You can access Sign In options, Share S4 with other people, select a light or dark them and gain access to our Support Portal.
You will also see options located along the bottom of the app for submitting new incidents, viewing a map that display incidents reported around you and an events options. The incidents near me and events do require you to log in with a valid account prior to accessing these features.
5. Submit an Incident
Submit An Incident
An incident can be reported from the S4 app whether you are logged in or accessing anonymously.
To report an incident, select the New Incident button located at the bottom left on the app.
Incident Information
When you select New Incident, you can provide details on the incident you are reporting. Choose from our common incident types and provide a detailed description of the incident.
The incident date and time will be auto selected, however these options are editable and you can choose a specific date and time that the incident occurred.
Incident Location/Permission to Contact
If you witnessed the incident in person, you can choose In Person for location. Alternatively, you can enter a specific address for the location where the incident occurred. If you are reporting an online incident, you can enter web addresses if needed.
The permissions section allows you to share the incident with the S4 community and you can select whether you grant permission to be contacted regarding this incident.
Add Persons and Vehicles
Descriptions for persons or vehicles associated with the incident you are reporting may be entered. Select the plus sign to add persons or vehicles, complete the fields that are relevant then tap the Done icon in the bottom right of the screen.
6. Attach Media
Attach Media
When submitting an incident, you can add various media types which include images, video and audio attachments.
To add these to the incident, tap the plus sign under Media Attachments and you will provided with the following options:
- Attach video from your phone library
- Record video from your phone camera
- Attach images from your phone library
- Capture images from your phone camera
- Record audio from your phone microphone
Once you have completed all the relevant fields, tap the Submit icon located in the bottom right screen.
7. My Incidents
My Incidents
If you have an S4 account and are signed in, you will see a list of all incidents you have submitted on the home screen of the app.
S4 (See Something Say Something) Administration Portal
1. Introduction
The S4 Admin Portal is a permissions based web application where S4 incidents submitted through the mobile app can be viewed and managed.
Users with access can view all incidents, download PDF reports of the incidents and manage an open/closed status for each incident.
In order to view the S4 Admin Portal, you must first create an account by clicking the Sign Up link on the log in page. Complete all the fields and if you allowed access, the appropriate permissions will be granted by Lumina administrators.
2. Incidents
S4 Incidents
When you access the S4 Admin Portal, all incidents that have been submitted will be displayed on the page. The newest incident will appear at the top of the list and you can view all details that were submitted.
This includes any media that were attached to the incident including images, videos, audio files and web URL’s.
3. Filtering Incidents
Filtering Incidents
You can choose to filter incidents based on the submitted date or by the open/closed status of the incident.
Simply choose a start and end date and a status to filter out incidents that meet your selection choices.
4. Manage Open / Closed Status
Open/Closed Status
After you have reviewed the incident and taken any necessary steps that may be required with law enforcement, you can choose to Close the incident.
Select the Close button and you will be prompted to confirm your selection. Each time an incident is close or reopened an automated note is generated for auditing purposes.
5. Notes
Add Notes
You can add notes to an incident as needed by selecting the plus sign located in the notes section.
This allows you to enter details of communications you may have had with law enforcement or other 3rd party organizations regarding the incident.
Each note entered will be visible in the notes section and a date/time stamp along with the email address of the person who entered the note will be displayed.
6. Media Attachments
View Media Attachments
If the incident was submitted with media attachments (images, video, audio or web links), you can view each of these on the incident.
Images are provided in a carousel for viewing on screen and videos may be viewed or downloaded to your local computer.
All media attachments are located beneath the incident summary information.
7. My Incidents
My Incidents
If you have an S4 account and are signed in, you will see a list of all incidents you have submitted on the home screen of the app.
8. Map
View Incident on Map
If the person submitting the incident includes an incident location, the address provided will be visible in the incident location along with a map icon.
Select the map icon to view details of the location on a map.
9. Download Report
Download Report
A PDF report can be downloaded for each incident. A download report button is located at the top of the incident, click this button and a PDF will be downloaded to your local computer or mobile device for viewing.