I found good features for classifying my data, but there are some noisy data that I cannot find any good features to exclude them, so I want to know if it is possible to have a class as 'others' (means the items in this class do not have any common properties but are different from other classes).
It seems you're facing a one-class-classification (OCC) problem, where you found "good" features to model your target class $C$ and now asking if it is possible to model the "outlier" class with other features, correct me if i'm right.
Under the assumption that your classification problem is an instance of OCC problems: It is not possible to model the outliers (what you refer to "others"). The reason is simply because you cannot model anything else besides $C$, due to absence of sample data from the "other" classes. In other words you are facing a recognition problem regarding $C$ rather than a discrimination problem between $C$ and the outlier class.
Try to find suitable (reliable) features for the target class $C$ in order to accept unseen objects as $C$ if these features are present and reject anything else where these features are not available. I hope you get the idea?