arXiv subjects

Dan Le

Publications and source records attributed to Dan Le.

3 recordsLinked to original sources

arXiv2020

ProtoQA: A Question Answering Dataset for Prototypical Common-Sense Reasoning

Given questions regarding some prototypical situation such as Name something that people usually do before they leave the house for work? a human can easily answer them via acquired experiences. There can be multiple right answers for such questions, with some more common for a situation than others. This paper introduces a new question answering dataset for training and evaluating common sense reasoning capabilities of artificial intelligence systems in such prototypical situations. The training set is gathered from an existing set of questions played in a long-running international game show FAMILY- FEUD. The hidden evaluation set is created by gathering answers for each question from 100 crowd-workers. We also propose a generative evaluation task where a model has to output a ranked list of answers, ideally covering all prototypical answers for a question. After presenting multiple competitive baseline models, we find that human performance still exceeds model scores on all evaluation metrics with a meaningful gap, supporting the challenging nature of the task.

cs.CL↗

arXiv2015

An Online Decoding Schedule Generating Algorithm for Successive Cancellation Decoder of Polar Codes

Successive cancellation (SC) is the first and widely known decoder of polar codes, which has received a lot of attentions recently. However, its decoding schedule generating algorithms are still primitive, which are not only complex but also offline. This paper proposes a simple and online algorithm to generate the decoding schedule of SC decoder. Firstly, the dependencies among likelihood ratios (LR) are explored, which lead to the discovery of a sharing factor. Secondly, based on the online calculation of the sharing factor, the proposed algorithm is presented, which is neither based on the depth-first traversal of the scheduling tree nor based on the recursive construction. As shown by the comparisons among the proposed algorithm and existed algorithms, the proposed algorithm has advantages of the online feature and the far less memory taken by the decoding schedule.

cs.IT↗

arXiv2014

Efficient Bit Sifting Scheme of Post-processing in Quantum Key Distribution

Bit sifting is an important step in the post-processing of Quantum Key Distribution (QKD) whose function is to sift out the undetected original keys. The communication traffic of bit sifting has essential impact on the net secure key rate of a practical QKD system, and it is facing unprecedented challenges with the fast increase of the repetition frequency of quantum channel. In this paper, we present an efficient bit sifting scheme whose core is a lossless source coding algorithm. Both theoretical analysis and experimental results demonstrate that the performance of our scheme is approaching the Shannon limit. Our scheme can greatly decrease the communication traffic of the post-processing of a QKD system, which means it can decrease the secure key consumption for classical channel authentication and increase the net secure key rate of the QKD system. Meanwhile, it can relieve the storage pressure of the system greatly, especially the device at Alice side. Some recommendations on the application of our scheme to some representative practical QKD systems are also provided.

quant-ph↗