Recent Submissions

  • How to initialize your network? Robust initialization for WeightNorm & ResNets 

    Arpit, Devansh; Campos Camúñez, Víctor; Bengio, Yoshua (2019)
    Conference lecture
    Open Access
    Residual networks (ResNet) and weight normalization play an important role in various deep learning applications. However, parameter initialization strategies have not been studied previously for weight normalized networks ...
  • Towards resilient EU HPC systems: A blueprint 

    Radojkovic, Petar; Marazakis, Manolis; Carpenter, Paul Matthew; Jeyapaul, Reiley; Gizopoulos, Dimitris; Schulz, Martin; Armejach Sanosa, Adrià; Ayguadé Parra, Eduard; Canal Corretger, Ramon; Moreto Planas, Miquel; Salami, Behzad; Unsal, Osman Sabri (2020-04)
    External research report
    Open Access
    This document aims to spearhead a Europe-wide discussion on HPC system resilience and to help the European HPC community define best practices for resilience. We analyse a wide range of state-of-the-art resilience mechanisms ...
  • Predictive autonomic transmission for low-cost low-margin metro optical networks 

    Ruiz Ramírez, Marc; Boitier, Fabien; Shariati, Mohammad Behnam; Layec, Patricia; Velasco Esteban, Luis Domingo (2020-10)
    Article
    Restricted access - publisher's policy
    Low-cost low-margin implementation plays an essential role in upgrading optical metro networks required for future 5G ecosystem. In this regard, low-resolution analog-to-digital converters can be used in coherent optical ...
  • Knowledge management in optical networks 

    Velasco Esteban, Luis Domingo; Tabatabaeimehr, Fatemehsadat; Ruiz Ramírez, Marc (Institute of Electrical and Electronics Engineers (IEEE), 2020)
    Conference report
    Open Access
    Autonomous network operation realized by means of control loops, where prediction from machine learning (ML) models is used as input to proactively reconfigure individual optical devices or the whole optical network, has ...
  • An experimental study of reduced-voltage operation in modern FPGAs for neural network acceleration 

    Salami, Behzad; Onural, Erhan Baturay; Yuksel, Ismail Emir; Koc, Fahrettin; Ergin, Oguz; Cristal Kestelman, Adrián; Unsal, Osman Sabri; Sarbazi-Azad, Hamid; Mutlu, Onur (Institute of Electrical and Electronics Engineers (IEEE), 2020)
    Conference report
    Open Access
    We empirically evaluate an undervolting technique, i.e., underscaling the circuit supply voltage below the nominal level, to improve the power-efficiency of Convolutional Neural Network (CNN) accelerators mapped to Field ...
  • POSTER: SPiDRE: accelerating sparse memory access patterns 

    Barredo Ferreira, Adrián; Beard, Jonathan C.; Moreto Planas, Miquel (Institute of Electrical and Electronics Engineers (IEEE), 2019)
    Conference lecture
    Open Access
    Development in process technology has led to an exponential increase in processor speed and memory capacity. However, memory latencies have not improved as dramatically and represent a well-known problem in computer ...
  • A novel FPGA-based high throughput accelerator for binary search trees 

    Melikoglu, Oyku; Ergin, Oguz; Salami, Behzad; Pavón Rivera, Julián; Unsal, Osman Sabri; Cristal Kestelman, Adrián (Institute of Electrical and Electronics Engineers (IEEE), 2019)
    Conference report
    Open Access
    This paper presents a deeply pipelined and massively parallel Binary Search Tree (BST) accelerator for Field Programmable Gate Arrays (FPGAs). Our design relies on the extremely parallel on-chip memory, or Block RAMs (BRAMs) ...
  • Distributed and centralized options for self-learning 

    Tabatabaeimehr, Fatemehsadat; Ruiz Ramírez, Marc; Velasco Esteban, Luis Domingo (2019)
    Conference report
    Open Access
    In general, the availability of enough real data from real fog computing scenarios to produce accurate Machine Learning (ML) models is rarely ensured since new equipment, techniques, etc., are continuously ...
  • Exceeding conservative limits: A consolidated analysis on modern hardware margins 

    Papadimitriou, George; Chatzidimitriou, Athanansios; Gizopoulos, Dimitris; Reddi, Vijay Janapa; Leng, Jingwen; Salami, Behzad; Unsal, Osman Sabri; Cristal Kestelman, Adrián (2020-06)
    Article
    Open Access
    Modern large-scale computing systems (data centers, supercomputers, cloud and edge setups and high-end cyber-physical systems) employ heterogeneous architectures that consist of multicore CPUs, general-purpose many-core ...
  • Soft-failure localization and time-dependent degradation detection for network diagnosis 

    Barzegar, Sima; Ruiz Ramírez, Marc; Velasco Esteban, Luis Domingo (2020)
    Conference report
    Open Access
    In optical networks, degradation of the Quality of Transmission (QoT) can be the outcome of soft-failures in optical devices, like Optical Transponders, Wavelength Selective Switches (WSS) and Optical ...
  • Aging-aware parallel execution 

    Medeiros, Thiarles S.; Berned, Gustavo P.; Navarro Muñoz, Antoni; Rossi, Fábio D.; Luizelli, Marcelo C.; Brandalero, Marcelo; Hübner, Michael; Beck, Antonio Carlos S.; Lorenzon, Arthur F. (Institute of Electrical and Electronics Engineers (IEEE), 2020-09-04)
    Article
    Open Access
    Computation has been pushed to the edge to decrease latency and alleviate the computational burden of the IoT applications in the cloud. However, the increasing processing demands of Edge Applications make necessary the ...
  • LoRaMoto: a communication system to provide safety awareness among civilians after an earthquake 

    Pueyo Centelles, Roger; Meseguer Pallarès, Roc; Freitag, Fèlix; Navarro Moldes, Leandro; Ochoa, Sergio; Santos, Rodrigo (Elsevier, 2021-02)
    Article
    Restricted access - publisher's policy
    Civilians in areas hit by earthquakes become in particular concerned about the safety of family and friends. However, outages in communications networks following these events limit their capacity to get information. So, ...

View more