Fast and accurate estimation of quantiles on data streams coming from communication networks, Internet Things (IoT), alike, is at the heart important processing applications including statistical analysis, latency monitoring, query optimization for parallel database management systems, more. Indeed, are more robust indicators underlying distribution, compared to moment-based such as mean variance. The streaming setting additionally constrains tracking quantiles, stream items may arrive a...

Accurate representation of oceanic conditions is fundamental for reliable climate modeling, weather forecasting, and environmental monitoring. However, ocean models observational datasets often exhibit systematic biases due to limitations in model physics, parameterizations, resolution, or coverage. In this work, we propose a diffusion bias correction. We systematically evaluated its performance Sea Surface Temperature on the sea surface temperature generation by varying different...

Summary Given an array of n elements and a value 2≤ k ≤ , frequent item or ‐majority element is occurring in more than / times. The problem requires finding all the elements. In this paper, we deal with parallel shared‐memory algorithms for items; present version Space Saving algorithm, study its behavior regard to accuracy performance on many multi‐core processors, including Intel Phi accelerator. We also investigate hybrid MPI/OpenMP against pure MPI‐based version. Through extensive...

Ship sailing is a complex endeavour, requiring carefully considered proactive and reactive strategies in choosing the course of action that best suits various events to be managed. Humans are already supported by different technologies for sailing, however these usually available isolation. In this paper we show how use simultaneously three fusing their information order provide enhanced support ship sailing. To our knowledge no similar approach reported literature from an operational point...

We deal with the problem of detecting frequent items in a stream under constraint that are weighted, and recent must be weighted more than older ones. This kind naturally arises wide class applications which data is considered useful valuable regard to older, stale data. The weight assigned an item is, therefore, function its arrival timestamp. As consequence, whilst traditional mining we need estimate frequency counts, instead required decayed counts. These said work time fading model. Two...

We present four CUDA based parallel implementations of the Space-Saving algorithm for determining frequent items on a GPU. The first variant exploits open-source CUB library to simplify implementation user's defined reduction, whilst second is our own reduction. third and fourth, built previous variants, are meant improve performance by taking advantage hardware atomic instructions. In particular, we implement warp ballot mechanism accelerate updates. show that coupled with update mechanism,...

Performance improvements in graphics hardware have made it possible to visualize complex virtual environments and provided opportunities interact with these a more realistic way. In this paper reality application for simulating billiards game is presented. A real billiard cue used its movements are reproduced the environment using visual marker detection system. No force feedback player. simulation has been built development OpenSceneGraph rigid body dynamics simulated utilizing PhysX...

The visualization of 3D models the patient’s body emerges as a priority in surgery. In this paper two different and interaction systems are presented: virtual interface low cost multi-touch screen. able to interpret real-time user’s movements can be used surgical pre-operative planning for navigation manipulation human built from CT images. surgeon visualize both standard patient information, such image dataset, model organs these developed is first prototype system designed avoid any...

We present UDDSketch (Uniform DDSketch), a novel sketch for fast and accurate tracking of quantiles in data streams. This is heavily inspired by the recently introduced DDSketch, based on bucket collapsing procedure that allows overcoming intrinsic limits corresponding DDSketch procedures. Indeed, does not allow derivation formal guarantees accuracy quantile estimation which follow sub-exponential distribution. On contrary, designed so can be given over full range arbitrary distribution...

A stream can be thought of as a very large set data, sometimes even infinite, which arrives sequentially and must processed without the possibility being stored. In fact, memory available to algorithm is limited it not possible store whole data instead scanned upon arrival summarized through succinct structure in order maintain only information interest. Two main tasks related processing are frequency estimation heavy hitter detection. The problem requires estimating each item, that number...

Performance improvements in graphics hardware and the diffusion of low cost haptic interfaces have made it possible to visualize complex virtual environments provided opportunities interact with these a more realistic way. In this paper Virtual Reality application game billiards is presented. By means commercial interface force feedback provided, thus rendering interaction exciting user. The introduction makes obtain simulation as strike billiard ball feel contact between cue ball....

We present parallel algorithms for mining Correlated Heavy Hitters from a two-dimensional data stream. In particular, we design and implement message-passing, shared-memory hybrid algorithm. To the best of our knowledge, these are first solving problem. show, through experimental results, that provide very good scalability, whilst retaining accuracy their sequential counterpart.