Outlier Detection with CIFAR10 Image Classifier
This demo is based on VAE outlier detection in the alibi detect project. Here we will :
- Launch an image classifier model trianed on the CIFAR10 dataset
- Setup an outlier detector for this particular model
- Send a request to get a image classification
- Send a perturbed request to get a outlier detection
NoteThis demo requires Knative installation on the cluster as the outlier detector will be installed as a kservice.
Create the model in its own dedicated namespace (needed for outlier detection). It should be a knative eventing namespace (labelled
knative-eventing-injection=enabled). For example as cluster admin for the namespace
kubectl create namespace outlier kubectl label namespace outlier knative-eventing-injection=enabled kubectl label namespace outlier seldon.restricted=false kubectl label namespace outlier serving.kubeflow.org/inferenceservice=enabled
Use the following model url
Configure the predictor log and set url as “http://default-broker":
Setup Request Logger
Go to ‘Setup Request Logger’ and enter:
Logger Name: seldon-request-logger Logger Image URI: docker.io/seldonio/seldon-request-logger:0.3.1
ELASTICSEARCH_HOST: elasticsearch-master.seldon-logs.svc.cluster.local ELASTICSEARCH_PORT: 9200 ELASTICSEARCH_PROTOCOL: http
Setup Outlier detector
Setup an outlier detector with model name
cifar10 using the default settings (which sets Reply URL as seldon-request-logger in current namespace) and storage URI as follows:
Run a single prediction using the tensorflow payload format of an image truck. Also a perturbed image of the truck in the same format at outlier truck image. Make a couple of these requests at random using the predict tool in the UI.
Monitor outliers on the Requests Screen
Go to the requests screen to view all the historical requests. You can see the outlier value on each instance. Also you can highlight outliers based on this score and also use the filter to see only the outliers as needed.
A dedicated namespace is recommended as knative eventing sends all requests to the outlier detector as well as the logger. In a shared namespace requests from other models would also go to the outlier detector, which is for cifar10.