Continuous maintenance of your model

Model validation

Model validation circle icon from humans in the loop with girl and robot with question on purple globe background

Once you have trained your initial ML model, you can use our services to verify its outputs in batches or in real time.

Once the model is running on real world data, a human in the loop can go through the model's predictions and correct false positives and false negatives which can afterwards be fed back for retraining

After reviewing your model's predictions, we can present you with an analysis of what classes the model has low performance on and we can gather additional data in order to increase it 

If you are not dealing with sensitive data or one that requires real-time processing but you still wish to improve your model's performance, you can use the same method but pass the data to us in batches:

You can set a threshold for the certainty of the model below which any data will be checked by a human in the loop

After going through the cases which are hard for your model, our humans in the loop can present you with insights and recommendations on how to improve its performance 

Once your model is up and running, you might want to pass the difficult and unusual cases that the model is not familiar with through a human:

You can set different thresholds for the model's certainty in order to determine how data is handled. For example, if the certainty is less than 50%, reroute the image to a human in the loop for a final decision 

We can we can commit a full-time or 24/7 team of humans in the loop to handle the cases where the certainty is low, especially in sensitive applications such as content moderation