A5039375454- Cloud Computing and Resource Management
- Parallel Computing and Optimization Techniques
- Distributed and Parallel Computing Systems
- Advanced Data Storage Technologies
- Software System Performance and Reliability
- Distributed systems and fault tolerance
- IoT and Edge/Fog Computing
- Peer-to-Peer Network Technologies
- Caching and Content Delivery
- Service-Oriented Architecture and Web Services
- Advanced Neural Network Applications
- Human Pose and Action Recognition
- Interconnection Networks and Systems
- Network Traffic and Congestion Control
- Hand Gesture Recognition Systems
- Mobile Agent-Based Network Management
- Data Management and Algorithms
- Video Analysis and Summarization
- Advanced Image and Video Retrieval Techniques
- Complex Network Analysis Techniques
- Multimodal Machine Learning Applications
- Hearing Impairment and Communication
- Speech and Audio Processing
- Network Security and Intrusion Detection
- Social Sciences and Policies
Universitat Politècnica de Catalunya
2015-2024
Barcelona Supercomputing Center
2015-2024
TU Dortmund University
2018
Istituto di Scienza e Tecnologie dell'Informazione "Alessandro Faedo"
2018
Massachusetts Institute of Technology
2018
University of California, Berkeley
2018
University of Nicosia
2018
Software (Spain)
2014
Universitat de Barcelona
2001-2009
First Technical University
2007-2009
In this work, we report the set-up and results of Liver Tumor Segmentation Benchmark (LiTS), which was organized in conjunction with IEEE International Symposium on Biomedical Imaging (ISBI) 2017 Conferences Medical Image Computing Computer-Assisted Intervention (MICCAI) 2018. The image dataset is diverse contains primary secondary tumors varied sizes appearances various lesion-to-background levels (hyper-/hypo-dense), created collaboration seven hospitals research institutions. Seventy-five...
We introduce SalGAN, a deep convolutional neural network for visual saliency prediction trained with adversarial examples. The first stage of the consists generator model whose weights are learned by back-propagation computed from binary cross entropy (BCE) loss over downsampled versions maps. resulting is processed discriminator to solve classification task between maps generated generative and ground truth ones. Our experiments show how training allows reaching state-of-the-art performance...
Interest has been growing in powering datacenters (at least partially) with renewable or "green" sources of energy, such as solar wind. However, it is challenging to use these because, unlike the "brown" (carbon-intensive) energy drawn from electrical grid, they are not always available. This means that demand and supply must be matched, if we take full advantage green minimize brown consumption. In this paper, investigate how manage a datacenter's computational workload match supply....
In this paper, we propose GreenSlot, a parallel batch job scheduler for datacenter powered by photovoltaic solar array and the electrical grid (as backup). GreenSlot predicts amount of energy that will be available in near future, schedules workload to maximize green consumption while meeting jobs' deadlines. If must used avoid deadline violations, selects times when it is cheap. Our results production scientific workloads demonstrate Green-Slot can increase up 117% decrease cost 39%,...
In this work, we report the set-up and results of Liver Tumor Segmentation Benchmark (LiTS), which was organized in conjunction with IEEE International Symposium on Biomedical Imaging (ISBI) 2017 Conferences Medical Image Computing Computer-Assisted Intervention (MICCAI) 2018. The image dataset is diverse contains primary secondary tumors varied sizes appearances various lesion-to-background levels (hyper-/hypo-dense), created collaboration seven hospitals research institutions. Seventy-five...
As energy-related costs have become a major economical factor for IT infrastructures and data-centers, companies the research community are being challenged to find better more efficient power-aware resource management strategies. There is growing interest in "Green" there still big gap this area be covered.
One of the factors that have hindered progress in areas sign language recognition, translation, and production is absence large annotated datasets. Towards this end, we introduce How2Sign, a multimodal multiview continuous American Sign Language (ASL) dataset, consisting parallel corpus more than 80 hours videos set corresponding modalities including speech, English transcripts, depth. A three-hour subset was further recorded Panoptic studio enabling detailed 3D pose estimation. To evaluate...
Cloud federation has been proposed as a new paradigm that allows providers to avoid the limitation of owning only restricted amount resources, which forces them reject customers when they have not enough local resources fulfill their customers' requirements. Federation provider dynamically outsource other in response demand variations. It also underused rent part providers. Both things could make get more profit used adequately. This requires clear understanding potential each decision,...
Recurrent Neural Networks (RNNs) continue to show outstanding performance in sequence modeling tasks. However, training RNNs on long sequences often face challenges like slow inference, vanishing gradients and difficulty capturing term dependencies. In backpropagation through time settings, these issues are tightly coupled with the large, sequential computational graph resulting from unfolding RNN time. We introduce Skip model which extends existing models by learning skip state updates...
The use of the Python programming language for scientific computing has been gaining momentum in last years. fact that it is compact and readable its complete set libraries are two important characteristics favour adoption. Nevertheless, still lacks a solution easily parallelizing generic scripts on distributed infrastructures, since current alternatives mostly require APIs message passing or restricted to embarrassingly parallel computations. In sense, this paper presents PyCOMPSs,...
Popular Internet services are hosted by multiple geographically distributed data centers. The location of the centers has a direct impact on services' response times, capital and operational costs, (indirect) carbon dioxide emissions. Selecting involves many important considerations, including its proximity to population centers, power plants, network backbones, source electricity in region, electricity, land, water prices at location, average temperatures location. As there can be potential...
We present a method for performing hierarchical object detection in images guided by deep reinforcement learning agent. The key idea is to focus on those parts of the image that contain richer information and zoom them. train an intelligent agent that, given window, capable deciding where attention among five different predefined region candidates (smaller windows). This procedure iterated providing analysis.We compare two candidate proposal strategies guide search: with without overlap....
The reduction of energy consumption in large-scale datacenters is being accomplished through an extensive use virtualization, which enables the consolidation multiple workloads a smaller number machines. Nevertheless, virtualization also incurs some additional overheads (e.g. virtual machine creation and migration) that can influence what best consolidated configuration, thus, they must be taken into account. In this paper, we present dynamic job scheduling policy for power-aware resource...
Molecular dynamics simulation (MD) is, just behind genomics, the bioinformatics tool that generates largest amounts of data, and is using amount CPU time in supercomputing centres. MD trajectories are obtained after months calculations, analysed situ, practice forgotten. Several projects to generate stable trajectory databases have been developed for proteins, but no equivalence exists nucleic acids world. We present here a novel database system store analyses acids. The initial data set...
Speech is a rich biometric signal that contains information about the identity, gender and emotional state of speaker. In this work, we explore its potential to generate face images speaker by conditioning Generative Adversarial Network (GAN) with raw speech input. We propose deep neural network trained from scratch in an end-to-end fashion, generating directly waveform without any additional identity (e.g reference image or one-hot encoding). Our model self-supervised approach exploiting...
We present a recurrent model for semantic instance segmentation that sequentially generates binary masks and their associated class probabilities every object in an image. Our proposed system is trainable end-to-end from input image to sequence of labeled and, compared methods relying on proposals, does not require post-processing steps its output. study the suitability our three different benchmarks, namely Pascal VOC 2012, CVPPP Plant Leaf Segmentation Cityscapes. Further, we analyze...
The growing complexity of software systems is resulting in an increasing number faults. According to the literature, faults are becoming one main sources unplanned system outages, and have important impact on company benefits image. For this reason, a lot techniques (such as clustering, fail-over techniques, or server redundancy) been proposed avoid failures, yet they still happen. Many failures those due aging phenomena. In work, we present detailed evaluation our chosen machine learning...
Spark has become one of the main options for large-scale analytics running on top shared-nothing clusters. This work aims to make a deep dive into parallelism configuration and shed light behavior parallel spark jobs. It is motivated by fact that application all available processors does not necessarily imply lower time, while may entail waste resources. We first propose analytical models expressing time as function number machines employed. then take another step, namely present novel...
We study the problem of dynamic resource allocation to clustered Web applications. extend application server middleware with ability automatically decide size clusters and their placement on physical machines. Unlike existing solutions, which focus maximizing utilization may unfairly treat some applications, approach introduced in this paper considers satisfaction each a particular attempts at least equally satisfy all model using utility functions, mapping CPU performance an relative its...