Daniel Etzold | Searcharxiv

arXiv2003

Improving spam filtering by combining Naive Bayes with simple k-nearest neighbor searches

Using naive Bayes for email classification has become very popular within the last few months. They are quite easy to implement and very efficient. In this paper we want to present empirical results of email classification using a combination of naive Bayes and k-nearest neighbor searches. Using this technique we show that the accuracy of a Bayes filter can be improved slightly for a high number of features and significantly for a small number of features.

cs.LG↗

arXiv2003

Heuristic to reduce the complexity of complete bipartite graphs to accelerate the search for maximum weighted matchings with small error

A maximum weighted matching for bipartite graphs $G=(A \cup B,E)$ can be found by using the algorithm of Edmonds and Karp with a Fibonacci Heap and a modified Dijkstra in $O(nm + n^2 \log{n})$ time where n is the number of nodes and m the number of edges. For the case that $|A|=|B|$ the number of edges is $n^2$ and therefore the complexity is $O(n^3)$. In this paper we want to present a simple heuristic method to reduce the number of edges of complete bipartite graphs $G=(A \cup B,E)$ with $|A|=|B|$ such that $m = n\log{n}$ and therefore the complexity of such that $m = n\log{n}$ and therefore the complexity of $O(n^2 \log{n})$. The weights of all edges in G must be uniformly distributed in [0,1].

cs.DS↗