Publications

Theano: Deep Learning on GPUs with Python
James Bergstra
Frédéric Bastien
Olivier Breuleux
Pascal Lamblin
Razvan Pascanu
Olivier Delalleau
Guillaume Desjardins
David Warde-Farley
Ian G Goodfellow
Arnaud Bergeron
Yoshua Bengio
In this paper, we present Theano 1 , a framework in the Python programming language for defining, optimizing and evaluating expressions invo… (voir plus)lving high-level operations on tensors. Theano offers most of NumPy’s functionality, but adds automatic symbolic differentiation, GPU support, and faster expression evaluation. Theano is a general mathematical tool, but it was developed with the goal of facilitating research in deep learning. The Deep Learning Tutorials 2 introduce recent advances in deep learning, and showcase how Theano
2011-12-31
(publié)
www.semanticscholar.org
Delayed Stochastic Decoding of LDPC Codes
Ali Naderi
Shie Mannor
Mohamad Sawan
Warren J. Gross
A new stochastic decoding algorithm, called Delayed Stochastic (DS) decoding, is introduced to implement low-density-parity-check (LDPC) dec… (voir plus)oders. The delayed stochastic decoding uses an alternative method to track probability values, which results in reduction of hardware complexity and memory requirement of the stochastic decoders. It is therefore suitable for fully-parallel implementation of long LDPC codes with applications in optical communications. Two decoders are implemented using the DS algorithm for medium (2048, 1723) and long (32768, 26624) LDPC codes. The decoders occupy 3.93- mm2 and 56.5- mm2 silicon area using 90-nm CMOS technology and provide maximum core throughputs of 172.4 and 477.7 Gb/s at [(Eb)/(No)]=5.5 and 4.8 dB, respectively.
2011-10-31
IEEE Transactions on Signal Processing (publié)
doi.org
Session details: Digital entertainment technologies and arts track posters
Mike Preuss
Christian Gagné
2011-07-11
Proceedings of the 13th annual conference companion on Genetic and evolutionary computation (publié)
doi.org
Session details: Digital entertainment technologies and arts track papers
Mike Preuss
Christian Gagné
2011-07-11
Proceedings of the 13th annual conference on Genetic and evolutionary computation (publié)
doi.org
Stochastic Multiple Stream Decoding of Cortex Codes
Matthieu Arzel
Cyril Lahuec
Christophe Jego
Warren J. Gross
Yvain Bruned
Being one of the most efficient solutions to implement forward error correction (FEC) decoders based on belief propagation, stochastic proce… (voir plus)ssing is thus a method worthy of consideration when addressing the decoding of emerging codes such as Cortex codes. This code family offers short block codes with large Hamming distances. Unfortunately, their construction introduces many hidden variables making them difficult to be efficiently decoded with digital circuits implementing the Sum-Product algorithm. With the introduction of multiple stochastic streams, the proposed solution alleviates the hidden variables problem thus yielding decoding performances close to optimal. Morevover, this new stochastic architecture is more efficient in terms of complexity-throughput ratio compared to recently published stochastic decoders using either edge or tracking forecast memories.
2011-06-30
IEEE Transactions on Signal Processing (publié)
doi.org
Stochastic Decoding of Turbo Codes
Quang Trung Dong
Matthieu Arzel
Christophe Jego
Warren J. Gross
Stochastic computation is a technique in which operations on probabilities are performed on random bit streams. Stochastic decoding of forwa… (voir plus)rd error-correction (FEC) codes is inspired by this technique. This paper extends the application of the stochastic decoding approach to the families of convolutional codes and turbo codes. It demonstrates that stochastic computation is a promising solution to improve the data throughput of turbo decoders with very simple implementations. Stochastic fully-parallel turbo decoders are shown to achieve the error correction performance of conventional a posteriori probability (APP) decoders. To our knowledge, this is the first stochastic turbo decoder which decodes a state-of-the-art turbo code. Additionally, an innovative systematic technique is proposed to cope with stochastic additions, responsible for the throughput bottleneck.
2010-11-30
IEEE Transactions on Signal Processing (publié)
doi.org
Relaxation Dynamics in Stochastic Iterative Decoders
Saeed Sharifi Tehrani
Chris Winstead
Warren J. Gross
Shie Mannor
Sheryl L. Howard
Vincent C. Gaudet
Stochastic decoding is a recently proposed approach for graph-based iterative error control decoding. We present and investigate three hyste… (voir plus)resis methods for stochastic decoding on graphs with cycles and show their close relationship with the successive relaxation method. Implementation results demonstrate the tradeoff in bit error rate performance with circuit complexity.
2010-10-31
IEEE Transactions on Signal Processing (publié)
doi.org
Greedy Gossip With Eavesdropping
Deniz Ustebay
Boris Oreshkin
Mark J. Coates
Michael G. Rabbat
This paper presents greedy gossip with eavesdropping (GGE), a novel randomized gossip algorithm for distributed computation of the average c… (voir plus)onsensus problem. In gossip algorithms, nodes in the network randomly communicate with their neighbors and exchange information iteratively. The algorithms are simple and decentralized, making them attractive for wireless network applications. In general, gossip algorithms are robust to unreliable wireless conditions and time varying network topologies. In this paper, we introduce GGE and demonstrate that greedy updates lead to rapid convergence. We do not require nodes to have any location information. Instead, greedy updates are made possible by exploiting the broadcast nature of wireless communications. During the operation of GGE, when a node decides to gossip, instead of choosing one of its neighbors at random, it makes a greedy selection, choosing the node which has the value most different from its own. In order to make this selection, nodes need to know their neighbors' values. Therefore, we assume that all transmissions are wireless broadcasts and nodes keep track of their neighbors' values by eavesdropping on their communications. We show that the convergence of GGE is guaranteed for connected network topologies. We also study the rates of convergence and illustrate, through theoretical bounds and numerical simulations, that GGE consistently outperforms randomized gossip and performs comparably to geographic gossip on moderate-sized random geometric graph topologies.
2010-06-30
IEEE Transactions on Signal Processing (publié)
doi.org
arxiv.org
Optimization and Analysis of Distributed Averaging With Short Node Memory
Boris Oreshkin
Mark J. Coates
Michael G. Rabbat
Distributed averaging describes a class of network algorithms for the decentralized computation of aggregate statistics. Initially, each nod… (voir plus)e has a scalar data value, and the goal is to compute the average of these values at every node (the so-called average consensus problem). Nodes iteratively exchange information with their neighbors and perform local updates until the value at every node converges to the initial network average. Much previous work has focused on algorithms where each node maintains and updates a single value; every time an update is performed, the previous value is forgotten. Convergence to the average consensus is achieved asymptotically. The convergence rate is fundamentally limited by network connectivity, and it can be prohibitively slow on topologies such as grids and random geometric graphs, even if the update rules are optimized. In this paper, we provide the first theoretical demonstration that adding a local prediction component to the update rule can significantly improve the convergence rate of distributed averaging algorithms. We focus on the case where the local predictor is a linear combination of the node's current and previous values (i.e., two memory taps), and our update rule computes a combination of the predictor and the usual weighted linear combination of values received from neighboring nodes. We derive the optimal mixing parameter for combining the predictor with the neighbors' values, and conduct a theoretical analysis of the improvement in convergence rate that can be achieved using this acceleration methodology. For a chain topology on N nodes, this leads to a factor of N improvement over standard consensus, and for a two-dimensional grid, our approach achieves a factor of ¿N improvement.
2010-04-30
IEEE Transactions on Signal Processing (publié)
doi.org
arxiv.org
Majority-Based Tracking Forecast Memories for Stochastic LDPC Decoding
Saeed Sharifi Tehrani
Ali Naderi
Guy-Armand Kamendje
Saied Hemati
Shie Mannor
Warren J Gross
2009-12-31
IEEE Transactions on Signal Processing (publié)
doi.org
Theano: A CPU and GPU Math Compiler in Python
James Bergstra
Olivier Breuleux
Frédéric Bastien
Pascal Lamblin
Razvan Pascanu
Guillaume Desjardins
Joseph Turian
David Warde-Farley
Yoshua Bengio
Theano is a compiler for mathematical expressions in Python that combines the convenience of NumPy's syntax with the speed of optimized nati… (voir plus)ve machine language. The user composes mathematical expressions in a high-level description that mimics NumPy's syntax and semantics, while being statically typed and functional (as opposed to imperative). These expressions allow Theano to provide symbolic differentiation. Before performing computation, Theano optimizes the choice of expressions, translates them into C++ (or CUDA for GPU), compiles them into dynamically loaded Python modules, all automatically. Common machine learn- ing algorithms implemented with Theano are from 1:6 to 7:5 faster than competitive alternatives (including those implemented with C/C++, NumPy/SciPy and MATLAB) when compiled for the CPU and between 6:5 and 44 faster when compiled for the GPU. This paper illustrates how to use Theano, outlines the scope of the compiler, provides benchmarks on both CPU and GPU processors, and explains its overall design.
2009-12-31
Proceedings of the Python in Science Conference (publié)
doi.org
Fully Parallel Stochastic LDPC Decoders
Saeed Sharifi Tehrani
Shie Mannor
Warren J. Gross
Stochastic decoding is a new approach to iterative decoding on graphs. This paper presents a hardware architecture for fully parallel stocha… (voir plus)stic low-density parity-check (LDPC) decoders. To obtain the characteristics of the proposed architecture, we apply this architecture to decode an irregular state-of-the-art (1056,528) LDPC code on a Xilinx Virtex-4 LX200 field-programmable gate-array (FPGA) device. The implemented decoder achieves a clock frequency of 222 MHz and a throughput of about 1.66 Gb/s at
2008-10-31
IEEE Transactions on Signal Processing (publié)
doi.org