Image reranking is an effective post-processing step to adjust the similarity order in image retrieval. As key components of initialized ranking lists, top-ranked neighborhoods of a given query usually play important ...Image reranking is an effective post-processing step to adjust the similarity order in image retrieval. As key components of initialized ranking lists, top-ranked neighborhoods of a given query usually play important roles in constructing dissimilarity measure. However, the number of pertinent candidates varies with respect to different queries. Thus the images with short lists of ground truth suffer from insufficient contextual information. It consequently introduces noises when using k-nearest neighbor rule to define the context. In order to alleviate this problem, this paper proposes auxiliary points which are added as assistant neighbors in an unsupervised manner. These extra points act on revealing implicit similarity in the metric space and clustering matched image pairs. By isometrically embedding each constructed metric space into the Euclidean space, the image relationships on underlying topological manifolds are locally represented by distance descriptions. Furthermore, by combining Jaccard index with our auxiliary points, we present a contextual modeling on auxiliary points ( CMAP ) method for image reranking. With richer contextual activations, the Jaccard similarity coefficient defined by local distribution achieves more reliable outputs as well as more stable parameters. Extensive experiments demonstrate the robustness and effectiveness of the proposed method.展开更多
Real-time performance is very important for recommender systems.In short video recommendation scenarios,users usually give explicit or implicit feedback in time during browsing,and the recommender system needs to sens...Real-time performance is very important for recommender systems.In short video recommendation scenarios,users usually give explicit or implicit feedback in time during browsing,and the recommender system needs to sense users'preferences in real time to meet their needs.However,traditional recommender systems are usually deployed on the cloud side,whenever the client requests the recommender system,it will return a list of short video results from the cloud side.Therefore,before the next recommendation request,the recommender system cannot adjust the recommendation result in real time according to the user's real-time feedback,resulting in an inaccurate recommender system on the cloud side.Consequently,in this paper,a cloud-edge joint strategy for short video recommendation(CloudEdgeRec)is proposed to address the aforementioned problems.Specifically,a lightweight model was deployed on edge devices to enable reranking based on user feedback.Furthermore,an interest-heuristic reranking(IHR)system was proposed to be implemented on the cloud side,which can provide a refresh mechanism to solve the problem that the limited cache on the edge devices cannot meet the drastic changes in user interests.The Markov decision process(MDP)is incorporated into IHR to preserve each generated distribution,and a matrix of exponential mean relevance is proposed to balance relationships between diversity and relevance.Finally,the experimental results show that both the offline evaluation of public datasets and online performance in short video platform demonstrate the effectiveness of CloudEdgeRec.展开更多
基金This work was supported in part by the Foundation for Innovative Research Groups of the National Natural Science Foundation of China (NSFC)(Grant No. 71421001)in part by the National Natural Science Foundation of China (NSFC)(Grant Nos. 61502073, 61772111 and 61429201)+1 种基金in part by the Fundamental Research Funds for the Central Universities (DUT18JC02)in part to Dr. Qi Tian by ARO (W911NF-15- 1-0290) and Faculty Research Gift Awards by NEC Laboratories of America and Blippar. This work was supported in part by the China Scholarship Council.
文摘Image reranking is an effective post-processing step to adjust the similarity order in image retrieval. As key components of initialized ranking lists, top-ranked neighborhoods of a given query usually play important roles in constructing dissimilarity measure. However, the number of pertinent candidates varies with respect to different queries. Thus the images with short lists of ground truth suffer from insufficient contextual information. It consequently introduces noises when using k-nearest neighbor rule to define the context. In order to alleviate this problem, this paper proposes auxiliary points which are added as assistant neighbors in an unsupervised manner. These extra points act on revealing implicit similarity in the metric space and clustering matched image pairs. By isometrically embedding each constructed metric space into the Euclidean space, the image relationships on underlying topological manifolds are locally represented by distance descriptions. Furthermore, by combining Jaccard index with our auxiliary points, we present a contextual modeling on auxiliary points ( CMAP ) method for image reranking. With richer contextual activations, the Jaccard similarity coefficient defined by local distribution achieves more reliable outputs as well as more stable parameters. Extensive experiments demonstrate the robustness and effectiveness of the proposed method.
文摘Real-time performance is very important for recommender systems.In short video recommendation scenarios,users usually give explicit or implicit feedback in time during browsing,and the recommender system needs to sense users'preferences in real time to meet their needs.However,traditional recommender systems are usually deployed on the cloud side,whenever the client requests the recommender system,it will return a list of short video results from the cloud side.Therefore,before the next recommendation request,the recommender system cannot adjust the recommendation result in real time according to the user's real-time feedback,resulting in an inaccurate recommender system on the cloud side.Consequently,in this paper,a cloud-edge joint strategy for short video recommendation(CloudEdgeRec)is proposed to address the aforementioned problems.Specifically,a lightweight model was deployed on edge devices to enable reranking based on user feedback.Furthermore,an interest-heuristic reranking(IHR)system was proposed to be implemented on the cloud side,which can provide a refresh mechanism to solve the problem that the limited cache on the edge devices cannot meet the drastic changes in user interests.The Markov decision process(MDP)is incorporated into IHR to preserve each generated distribution,and a matrix of exponential mean relevance is proposed to balance relationships between diversity and relevance.Finally,the experimental results show that both the offline evaluation of public datasets and online performance in short video platform demonstrate the effectiveness of CloudEdgeRec.
