École Polytechnique de Montréal

About: École Polytechnique de Montréal is a education organization based out in Montreal, Quebec, Canada. It is known for research contribution in the topics: Finite element method & Computer science. The organization has 8015 authors who have published 18390 publications receiving 494372 citations.

Zoneout: Regularizing RNNs by Randomly Preserving Hidden Activations

Abstract: We propose zoneout, a novel method for regularizing RNNs At each timestep, zoneout stochastically forces some hidden units to maintain their previous values Like dropout, zoneout uses random noise to train a pseudo-ensemble, improving generalization But by preserving instead of dropping hidden units, gradient information and state information are more readily propagated through time, as in feedforward stochastic depth networks We perform an empirical investigation of various RNN regularizers, and find that zoneout gives significant performance improvements across tasks We achieve competitive results with relatively simple models in character- and word-level language modelling on the Penn Treebank and Text8 datasets, and combining with recurrent batch normalization yields state-of-the-art results on permuted sequential MNIST

Phase and amplitude sensitivities in surface plasmon resonance bio and chemical sensing

Abstract: We consider amplitude and phase characteristics of light reflected under the Surface Plasmon Resonance (SPR) conditions and study their sensitivities to refractive index changes associated with biological and chemical sensing. Our analysis shows that phase can provide at least two orders of magnitude better detection limit due to the following reasons: (i) Maximal phase changes occur in the very dip of the SPR curve where the vector of probing electric field is maximal, whereas maximal amplitude changes are observed on the resonance slopes: this provides a one order of magnitude larger sensitivity of phase to refractive index variations; (ii) Under a proper design of a detection scheme, phase noises can be orders of magnitude lower compared to amplitude ones, which results in a much better signal-to-noise ratio; (iii) Phase offers much better possibilities for signal averaging and filtering, as well as for image treatment. Applying a phase-sensitive SPR polarimetry scheme and using gas calibration model, we experimentally demonstrate the detection limit of 10(-8) RIU, which is about two orders of magnitude better compared to amplitude-sensitive schemes. Finally, we show how phase can be employed for filtering and treatment of images in order to improve signal-to-noise ratio even in relatively noisy detection schemes. Combining a much better physical sensitivity and a possibility of imaging and sensing in micro-arrays, phase-sensitive methodologies promise a substantial upgrade of currently available SPR technology.