A machine learning model has been trained to recognize the name of Animals and Plants. If suppose an automobile name is given, is it possible to say that the given name doesn't belong to the category animals or plants. If possible, kindly mention the methodology or algorithm which achieves this scenario.
E.g. If 'Lion' or 'Coconut Tree' is given the model will be predicting either 'Animals' or 'Trees' category. If suppose, 'Audi' is given, is it possible to say that the given item belongs neither to 'Animals' or 'Plants'. (Note : I have heard that the machine learning model will try to fit into either one the category).
A classifier actually gives you a probability of item belonging to a category, unless you add a final layer or post-processing that translates those probabilities to one and zeros. So, you can define a certain confidence threshold for probabilities and if classifier does not output probabilities above the threshold then call the output undecided.
An "audi" can still have features that make network believe it is tree for example.