Lifelogging aims to capture a person’s life experiences using digital devices. When captured over an extended period of time a lifelog can potentially contain millions of files from various sources in a range of formats. For lifelogs containing such massive numbers of items, we believe it is important to group them into meaningful sets and summarize them, so that users can search and browse their lifelog data efficiently. Existing studies have explored the segmentation of continuously captured images over short periods of at most a few days into small groups of
“events” (episodes). Yet, for long-term lifelogs, higher levels of abstraction are desirable due to the very large number of “events” which will occur over an extended period. We aim to segment a long-term lifelog at the level of general events which typically extend beyond a daily boundary, and to select summary information to represent these events. We describe our current work on higher level segmentation and summary information extraction for long term life logs and report a preliminary pilot study on a real long-term lifelog collection.
Cellular traffic prediction enables operators to adapt to traffic demand in real-time for improving network resource utilization and user experience. To predict cellular traffic, previous studies either applied Recurrent Neural Networks (RNN) at individual base stations or adapted Convolutional Neural Networks (CNN) to work at grid-cells in a geographically defined grid. These solutions do not consider explicitly the effect of handover on the spatial characteristics of the traffic, which may lead to lower prediction accuracy. Furthermore, RNN solutions are slow to train, and CNN-grid solutions do not work for cells and are difficult to apply to base stations. This paper proposes a new prediction model, STGCN-HO, that uses the transition probability matrix of the handover graph to improve traffic prediction. STGCN-HO builds a stacked residual neural network structure incorporating graph convolutions and gated linear units to capture both spatial and temporal aspects of the traffic. Unlike RNN, STGCN-HO is fast to train and simultaneously predicts traffic demand for all base stations based on the information gathered from the whole graph. Unlike CNN-grid, STGCN-HO can make predictions not only for base stations, but also for cells within base stations. Experiments using data from a large cellular network operator demonstrate that our model outperforms existing solutions in terms of prediction accuracy.
We are delighted to welcome you to SIGMOD 2012, the 2012 edition of the ACM SIGMOD International Conference on Management of in Scottsdale, Arizona, in the Southwest of the United States. Scottsdale is in the heart of the Sonoran Desert and offers stunning desert vistas and a breathtaking setting for the conference. At the same time, Scottsdale is adjacent to Phoenix, one of the largest and fastest-growing cities in the United States.
SIGMOD 2012 hosts an exciting technical program, with two keynote talks, by Pat Hanrahan (Stanford University and Tableau Software) and Amin Vahdat (University of California, San Diego and Google); a plenary session with Perspectives on Big Data, by Donald Kossmann (ETHZ), Kristen LeFevre (Google Research and University of Michigan), Sam Madden (MIT), and Anand Rajaraman (@WalmartLabs); 48 research paper presentations; six tutorials; 30 demonstrations; and 18 industrial presentations. In addition to having full 30-minute presentation slots, research papers are included in one of two Research Plenary Poster Sessions. One of these sessions is jointly for PODS and SIGMOD research papers, to deepen the ties between the two conferences. Another new plenary poster session, for papers from the 11 workshops co-located with SIGMOD 2012, is an effort to strengthen the link and synergy between the workshops and the conference.
SIGMOD 2012 includes several technical and social events designed specifically for student attendees. The SIGMOD/PODS 2012 Ph.D. Symposium, the Database Mentoring Workshop, the Undergraduate Research Poster Competition, and the New Researcher Symposium are all established components of the SIGMOD program and are all part of SIGMOD 2012. The conference also hosts a session dedicated to highlighting the finalists of the SIGMOD Programming Contest. (This year's task is to implement a multidimensional, high-throughput, in-memory indexing system.) In addition, the conference includes a new Information Session on Careers in Industry, aimed at bringing student attendees together with our Gold, Platinum, and Diamond sponsors, as well as vis-a-vis meetings aimed at helping Ph.D. students meet internationally recognized researchers in their research areas, to exchange ideas and receive guidance in a relaxed social setting.
While approaching of the new century,Great changes in universties have taken place in Physical Educational aim,task,teaching mode,strategic status,P.E. will be more open and pefect,teaching material and method will be more civilized.Educational goal is going to develop from pure body education to lifelong education.Therefore,phyical educational teachers in unversities is demanded to reach higher more overall qualities.
In this article, we present a multi-class blue noise sampling algorithm by throwing samples as the constrained Wasserstein barycenter of multiple density distributions. Using an entropic regularization term, a constrained transport plan in the optimal transport problem is provided to break the partition required by the previous Capacity-Constrained Voronoi Tessellation method. The entropic regularization term cannot only control spatial regularity of blue noise sampling, but it also reduces conflicts between the desired centroids of Vornoi cells for multi-class sampling. Moreover, the adaptive blue noise property is guaranteed for each individual class, as well as their combined class. Our method can be easily extended to multi-class sampling on a point set surface. We also demonstrate applications in object distribution and color stippling.
D (2-dimension) bar code has many advantages compared with 1-D bar code, so it has many prevalent applications in recent days. However, there're several factors that affecting the sampling process, they lead to a certain degree of deformation towards the 2-D bar code. Furthermore, this deformation may too severe for 2-D bar code to be recognized. Generally, the main stream of twisted image correcting is based on the approach of control-point transformation. So how to locate the control point accurately becomes a key factor in 2-D bar code correcting. This paper develops a QR barcode searching graph and gets the precise location of control point. Experiment shows that our algorithm has good efficiency and accuracy.
Adverse drug events (ADEs) are a serious health problem that can be life-threatening. While a lot of work on detecting correlation between a drug and an ADE, limited studies have been conducted on personalized ADE risk prediction. Avoiding the drugs with high likelihood of causing severe ADEs helps physicians to provide safer treatments to patients. The goal of this study is to assess personalized ADE risks that a target drug may induce on a target patient, based on patient medical history recorded in claim codes, which provide information about diagnosis, drugs taken, related medical supplies besides billing information. We developed a HTNNR model (Hierarchical Time-aware Neural Network for ADE Risk) that captures characteristics of claim codes and their relationship. Eempirical evaluation shows that the proposed HTNNR model substantially outperforms the comparison methods.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
