Recently, locality-constrained linear coding (LLC) as a coding strategy has attracted much attention, due to its better reconstruction than sparse coding and vector quantization. However, LLC ignores the weight information of codewords during the coding stage, and assumes that every selected base has same credibility, even if their weights are different. To further improve the discriminative power of LLC code, we propose a weighted LLC algorithm that considers the codeword weight information. Experiments on the KTH and UCF datasets show that the recognition system based on WLLC achieves better performance than that based on the classical LLC and VQ, and outperforms the recent classical systems.

Under random linear coding (RLC) scheme, we present a simple expression of the probability distribution p(D=K) for decoding delay D incurred by the lossy channel, where K is a positive integer. In contrast with the previous contribution, our focus is firstly on deriving the cumulative distribution function of the discrete random variable D over a perfect channel. One benefit of such dispose is that, from the overall viewpoint, computing the cumulative distribution function of delay D can be related with calculating the cardinalities of sets of some special matrices, so that the former can be obtained from the latter. Moreover, our expression of the probability distribution is an explicit form, and is valid for any number of packets M, freewill field size q and arbitrary channel loss rate ε.

We show the sufficient conditions for coding by nodes such that every sink can decode all information multicasted by the single source when there exists time-varying delay of information transmission at links in the network coding.