27 September 2019 in IEEE Transactions on Cloud Computing
Cloud storage has become a primary industry in remote data management service but also attracts security concerns, where the best available approach for preventing data disclosure is encryption. Among them the public key encryption with keyword search (PKSE) is considered to be a promising technique, since clients can efficiently search over encrypted data files. That is, a client first generates a search token when to query data files, the cloud server uses the search token to proceed the query over encrypted data files. However, a serious attack is raised when PKSE meets cloud. Formally speaking, the cloud server can learn the information of a newly added encrypted data file containing the keyword that previously queried by using the search tokens it has received, and can further discover the privacy information. To address this issue, we propose a forward secure public key searchable encryption scheme, in which a cloud server cannot learn any information about a newly added encrypted data file containing the keyword that previously queried. To better understand the design principle, we introduce a framework for constructing forward secure public key searchable encryption schemes based on attribute-based searchable encryption. Finally, the experiments show our scheme is efficient.
20 September 2022 in IEEE Transactions on Artificial Intelligence
Current neural language models are trained to minimize cross-entropy and use softmax to compute the locally normalized probabilities over the target. While this setup provides solid results in several natural language processing (NLP) tasks, one unsatisfying aspect is its dense output. This density is wasteful, making models hard to interpret and assigning probability mass to many implausible outputs. To overcome this problem, we propose T-softmax, a simple but effective method to draw considerably sparse probability out of neural language models than softmax. Our method avoids dense output by truncating the unreliable tail of the probability distribution to improve the model's performance. In addition, we generalize logits with temperature, a critical regularization technique, from the softmax to T-softmax. To show our approach as a drop-in replacement for softmax, we evaluate them on three NLP tasks: summary generation, question answer, and math word problem. Experimental results show that our proposed model significantly improves performance without sacrificing speed; notably, in all experiments, our method outperforms the softmax.