This paper shows that for the multiple distribution estimation filter (MDEF)[1] [2], the one-step-behind (OSB) smoothed state probability density function (PDF) used for the estimation of the filtered state PDF is the key factor for the high filtering accuracy of the MDEF. The MDEF calculates the OSB smoothed state PDF prior to the calculation of the filtered state PDF at the current epoch and the filtered state PDF is estimated by the marginalization of the conditional state PDF with respect to this smoothed state PDF. Since the OSB smoothed state PDFs are obtained at every time step prior to the observation update, the MDEF can be regarded as providing the simultaneous estimation of the filtered and OSB smoothed state PDFs. In this paper, we numerically evaluate the estimation accuracy for the OSB smoothed state estimates by the MDEF using the benchmark filtering problems [3]-[5] and compare it with those for the particle and the Gaussian smoothers employed to the MDEF. We confirmed that the smoothing accuracy for the OSB smoothed state estimates was more accurate than that for the Gaussian smoother and almost comparable to that for the particle smoother, while the calculation cost was significantly lowered than that for the particle smoother.

We adopt the generalized Pareto distribution for the information-based model and show that the parameters can be estimated based on the mean excess function. The proposed information retrieval model corresponds to the extension of the divergence from independence and is designed to be data-driven. The proposed model is then applied to the specific object search called the instance search and the effectiveness is experimentally confirmed.

This paper deals with a novel concept of an exponential IDF in the BM25 formulation and compares the search accuracy with that of the BM25 with the original IDF in a content-based video retrieval (CBVR) task. Our video retrieval method is based on a bag of keypoints (local visual features) and the exponential IDF estimates the keypoint importance weights more accurately than the original IDF. The exponential IDF is capable of suppressing the keypoints from frequently occurring background objects in videos, and we found that this effect is essential for achieving improved search accuracy in CBVR. Our proposed method is especially designed to tackle instance video search, one of the CBVR tasks, and we demonstrate its effectiveness in significantly enhancing the instance search accuracy using the TRECVID2012 video retrieval dataset.

Information needs expressed by using the same query for a search engine might be totally different, whether on week days or weekends, or during the day or at night. For queries having no temporal changes in search intentions, the same search results ranking may be returned regardless of the time, but for those with temporal changes the ranking must be suitably altered depending on the time of input. To achieve time-dependent search results rankings, we focus on the temporal changes in the search intentions. We present the results obtained by analyzing a commercial search engine log and propose a method of detecting queries showing periodic changes in the search intentions.