Afaik, the problem with many machine learning algorithms is that they will often label nonsense into some categories.
What measures can one take to discard nonsense results?
Eg. if you have a bot that is supposed to label stuff as either orange or an apple, how can you gurantee that it won't label a random image of noise as either of the two?
Or label a car as either of the two?
Do you have to train your neural network to have 3 categories, "apple", "orange" and "neither"?
The problem you're describing can perhaps be treated as "anomaly detection". Google the term in quotes, you'll get a ton of references. So the idea would be to first train an anomaly detector to separate the "apples, oranges, and other fruits" from non-fruits (or "nonsense" in OP terminology). Then train a classifier on the fruits.
Update. In light of additional answers -- and too long for a comment. The problem with declaring a "non-fruit" category highlights the distinction between classification and anomaly detection. In classification, we assume 2 or more coherent classes. Apples, oranges, bananas -- these all have well-defined common characteristics. On the other hand, "non-fruit" is not a coherent category. It can only be meaningfully defined as something like "not fulfilling enough of the fruit characteristics". Anomalous objects are not a coherent class; they are by definition anything that is not sufficiently close to a "normal" object. We do not expect examples of anomalous objects to be informative for training. After all, you can just generate random pixel-images and get tons of anomalous examples for free!
This is why anomaly detection requires techniques distinct from classification.
Do you have to train your neural network to have 3 categories, "apple", "orange" and "neither"?
Yes, you would have to.
Eg. if you have a bot that is supposed to label stuff as either orange or an apple, how can you gurantee that it won't label a random image of noise as either of the two?
You would need to include images of the third category "neither-apple-orange" so that you train your classifier to learn three categories.
