Performance calibration allows you to test, fine-tune, and simulate running event detection models using continuous real-world or synthetically generated streams of data. It is designed to provide an immediate understanding of how your model is expected to perform in the field.
Currently only available for Audio data projects
Performance calibration is currently only available for projects that contain audio data. It's designed for use with projects that are detecting specific events (such as spoken keywords), as opposed to classifying ambient conditions. Please stay tuned for future information on support for other types of sensor data!
Performance Calibration is a tool for testing and configuring embedded machine learning pipelines for event detection. It provides insight into how your pipeline will perform on streaming data, which is what your application will encounter in the real world. It works within Studio, and does not require you to deploy to a physical device.
After testing is complete, you can use Performance Calibration to configure a post-processing algorithm that will interpret the output of your ML pipeline, transforming it into a stream of actionable events. The results of testing are used to help guide selection of the optimal post-processing algorithm for your use case.
For example, a developer working on a keyword spotting application could use Performance Calibration to understand how well their ML pipeline detects keywords in a sample of real world audio, and to select the post-processing algorithm that provides the best quality output.
Performance Calibration gives you an accurate prediction of how your ML pipeline will perform when it is deployed in the real world. Analyzing real world performance before deployment in the field allows you to iterate on your pipeline much more quickly, helping you identify and solve common performance issues much earlier in the process.
Interpreting the output of an ML pipeline on streaming data requires a post-processing algorithm, which edge ML developers have traditionally had to write and tune by hand, balancing the trade-off between false positives and false negatives to fit their particular use case. By quantifying and automating this process, Performance Calibration gives developers precise control over the trade-offs they select for their application.
Performance can be measured using either recordings of real-world data, or with realistic synthetic recordings generated using samples from your test dataset. This allows you to easily test your model’s performance under various scenarios, such as varying levels of background noise, or with different environmental sounds that might occur in your deployment environment.
When Performance Calibration runs, your ML pipeline is run across the input data with the same latency as is predicted for the target selected on the Dashboard page of your project. This results in a set of raw predictions which must be filtered by a post-processing algorithm to produce a signal every time a particular event class is detected.
The post-processing algorithm has configurable parameters that determine the overall performance of the pipeline. These parameters can be adjusted to control the trade-off between false acceptance rate (how often an event is mistakenly detected) and false rejection rate (how often an event is mistakenly ignored). This allows you to determine how sensitive your application is to inputs.
False positives and false negatives
No ML model is perfect, so developers using ML for event detection always need to pick a trade-off between false positives and false negatives. The appropriate trade-off depends on the application. For example, if you're attempting to detect a dangerous situation in an industrial facility, it may be important to minimize false negatives. On the other hand, if you're concerned about annoying users with unintentional activations of a smart home device, you may wish to minimize false positives.
The following page walks through the process of using Performance Calibration with an example project. Check out our blog post for more information!
First, make sure you have an audio project in your Edge Impulse account. No projects yet? Follow one of our tutorials to get started:
Or, clone the "Bird sound classifier" project that is used in this documentation to your Edge Impulse account: https://studio.edgeimpulse.com/public/16060/latest
Once you've trained your impulse, select the Performance calibration tab and set your testing configuration settings:
Select noise labels. Which label is used to represent generic background noise or "silence"?
Select any other labels that should be ignored by your application, i.e. other classes that are equivalent to background noise or "silence".
Choose an audio sample type: simulated real world audio or upload your own in a zip file.
Then, click Run test.
Simulated real world audio is a synthetically generated audio stream consisting of samples taken from your testing dataset and layered on top of artificial background noise. For free Edge Impulse projects, you can choose to generate either 10 minutes or 30 minutes of simulated real world audio.
Already have a long, real-world recording of background noise which includes your target model's classes? Upload your own audio sample (.wav) in a zip file, along with its Label Tracks in Audacity format (.txt).
Your impulse can be configured with a post-processing algorithm that will minimize either false activations or false rejections. The chart shows a range of suggested configs. If you save one, it will be used when your impulse is deployed.
Selecting from the various "Suggested config" icons on the FRR/FAR chart will update the Selected config information. Click on Save selected config to use the selected FAR and FRR trade off when your impulse is deployed. This config information is also accessible in the deployed Edge Impulse library.
Mean FAR: The mean False Acceptance Rate. Measures how often labels are mistakenly detected. Does not include statistics for noise labels.
Mean FRR: The mean False Rejection Rate. Measures how often events are mistakenly missed. Does not include statistics for noise labels.
Averaging window duration (ms): The raw inference results are averaged across this length of time.
Detection threshold (ms): A class is considered a positive match when it exceeds this threshold.
Suppression period (ms): Matches are ignored for this length of time following a positive result.
Shows the performance statistics for each label.
FAR: False Acceptance Rate. Measures how often a label is mistakenly detected.
FRR: False Rejection Rate. Measures how often a label is mistakenly missed.
True Positives: The number of times each label was correctly triggered.
False Positives: The number of times each label was incorrectly triggered.
True Negatives: The number of times each label was correctly not triggered.
False Negatives: The number of times each label was incorrectly not triggered.
False acceptance rate and false rejection rate
FAR is also sometimes known as the False Positive Rate, and FRR as the False Negative Rate. These industry-standard metrics are calculated as follows:
Shows any errors your impulse makes on a sample of data, with a table of results.
Error: False positives are displayed in red while false negatives are displayed in blue.
Type: Spurious match, incorrect match, duplicate match, or blank.
Label: The data label the model predicted in the audio stream.
Start time: The timestamp starting location of the selected error in the audio data stream.
Play button: Preview the audio stream at the error's start time.
What we refer to as "Ground Truth" in this context is the sound/label association that the synthetically generated audio contains at a given time.
Incorrect match: A detection matches the wrong ground truth
Spurious match: This match detection has not been associated with any ground truth.
Duplicate match: The same ground truth was detected more than once. The first correct detection is considered a true positive but subsequent detections are considered false positives.
