Bheemarjuna Reddy Tamma

The paper proposes SEED, a novel semantic-structure-agnostic semi-supervised continual learning method that significantly improves malware detection performance under limited labeling by leveraging representation consistency and uncertainty quantification.