1 paper · 1 filter
Filip Graliński, Ryszard Staruch, Krzysztof Jurkiewicz
We present a new method to detect anomalies in texts (in general: in sequences of any data), using language models, in a totally unsupervised manner. The method considers probabili…