An Android mutation malware detection based on deep learning using visualization of importance from codes

摘要: Abstract Smartphone use, especially the Android platform, has already got 80% market shares, due to an aforementioned [where?] report, it becomes attacker's primary objective. There is a growing number of storing private data onto smart phones and low safety defense measures, attackers can use multiple ways launch attack user's smartphones. (e.g. Using different coding style confuse malware detecting software). Existing detection methods features, like sensor API, system call, control flow structure information flow, then also machine learning check whether its or not. These features provide app's unique property limitation, that say, from some perspectives might suit for specific attack, but wouldn't others. Nowadays most only one these mostly analyze detect code, facing code confusion zero-day attacks, feature's extraction method may cause wrong judgement. So, it's necessary design effective technique analysis prevent malware. In this paper, we importance words apk, because confusion, rename variables. If using general static cannot judge correctly, values go through our proposed generate image, finally convolutional neural network decide apk file