In this paper we present the techniques used for the University of Montréal's team submissions to the 2013 Emotion Recognition in the Wild Challenge. The challenge is to classify the emotions expressed by the primary human subject in short video clips extracted from feature length movies. This involves the analysis of video clips of acted scenes lasting approximately one-two seconds, including the audio track which may contain human voices as well as background music. Our approach combines multiple deep neural networks for different data modalities, including: (1) a deep convolutional neural network for the analysis of facial expressions within video frames; (2) a deep belief net to capture audio information; (3) a deep autoencoder to model the spatiotemporal information produced by the human actions depicted within the entire scene; and (4) a shallow network architecture focused on extracted features of the mouth of the primary human subject in the scene. We discuss each of these techniques, their performance characteristics and different strategies to aggregate their predictions. Our best single model was a convolutional neural network trained to predict emotions from static frames using two large data sets, the Toronto Face Database and our own set of faces images harvested from Google image search, followed by a per frame aggregation strategy that used the challenge training data. This yielded a test set accuracy of 35.58%. Using our best strategy for aggregating our top performing models into a single predictor we were able to produce an accuracy of 41.03% on the challenge test set. These compare favorably to the challenge baseline test set accuracy of 27.56%.

We show the emergence of S-RNA under stretching in analogy to S-DNA and propose a method for its detection using electrical measurement.

In this study, we compare the charge transport properties of multiple (double stranded) dsRNA sequences with corresponding dsDNA sequences. Recent studies have presented a contradictory picture of relative charge transport...

The DNA molecule, apart from carrying the genetic information, plays a crucial role in a variety of biological processes and finds applications in drug design, nanotechnology and nanoelectronics. The molecule undergoes significant structural transitions under the influence of forces due to physiological and non-physiological environments. Here, we summarize the insights gained from simulations and single-molecule experiments on the structural transitions and mechanics of DNA under force, as well as its elastic properties, in various environmental conditions, and discuss appealing future directions.

Charge transport in deoxyribonucleic acid (DNA) is of immense interest in biology and molecular electronics. Electronic coupling between the DNA bases is an important parameter describing the efficiency of charge transport in DNA. A reasonable estimation of this electronic coupling requires many expensive first principle calculations. In this article, we present a machine learning (ML) based model to calculate the electronic coupling between the guanine bases of the DNA (in the same strand) of any length, thus avoiding expensive first-principle calculations. The electronic coupling between the bases are evaluated using density functional theory (DFT) calculations with the morphologies derived from fully atomistic molecular dynamics (MD) simulations. A new and simple protocol based on the coarse-grained model of the DNA has been used to extract the feature vectors for the DNA bases. A deep neural network (NN) is trained with the feature vector as input and the DFT-calculated electronic coupling as output. Once well trained, the NN can predict the DFT-calculated electronic coupling of new structures with a mean absolute error (MAE) of 0.02 eV.