A visual attention system, inspired by the behavior and the neuronal architecture of the early primate visual system, is presented. Multiscale image features are combined into a single topographical saliency map. A dynamical neural network then selects attended locations in order of decreasing saliency. The system breaks down the complex problem of scene understanding by rapidly selecting, in a computationally efficient manner, conspicuous locations to be analyzed in detail.

/pdf/a-model-of-saliency-based-visual-attention-for-rapid-scene-wv0n3yqxix.pdf

A model of saliency-based visual attention for rapid scene analysis

Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

Data Mining Practical Machine Learning Tools and Techniques

This paper presents a method for recognizing scene categories based on approximate global geometric correspondence. This technique works by partitioning the image into increasingly fine sub-regions and computing histograms of local features found inside each sub-region. The resulting "spatial pyramid" is a simple and computationally efficient extension of an orderless bag-of-features image representation, and it shows significantly improved performance on challenging scene categorization tasks. Specifically, our proposed method exceeds the state of the art on the Caltech-101 database and achieves high accuracy on a large database of fifteen natural scene categories. The spatial pyramid framework also offers insights into the success of several recently proposed image descriptions, including Torralbas "gist" and Lowes SIFT descriptors.

/pdf/beyond-bags-of-features-spatial-pyramid-matching-for-4aa5cbe2fz.pdf

Beyond Bags of Features: Spatial Pyramid Matching for Recognizing Natural Scene Categories

The information-transfer rate (ITR) is commonly used to assess the performance of brain-computer interfaces (BCI). Various studies have shown that the optimal number of mental tasks to be used is fairly low, around 3 or 4. We propose a formal approach and an experimental validation to demonstrate and confirm that this optimum is user and BCI design dependent. Even if increasing the number of mental tasks to the optimum indeed leads to an increase of the ITR, the gain remains small. This might not justify the added complexity in terms of protocol design.

/pdf/information-transfer-rate-modeling-of-eeg-based-synchronized-43i1xxjwo0.pdf

Information-transfer rate modeling of EEG-based synchronized brain-computer interfaces

Purpose – The purpose of this paper is to overcome the limitations of sensory substitution methods (SSDs) to represent high-level or conceptual information involved in vision, which are mainly produced by the biological sensory mismatch between sight and substituting senses. Thus, provide the visually impaired with a more practical and functional SSD. Design/methodology/approach – Unlike any other approach, the SSD extends beyond a sensing prototype, by integrating computer vision methods to produce reliable knowledge about the physical world (at the lowest cost to the user). Importantly though, the authors do not abandon the typical encoding of low-level features into sound. The paper simply argues that any visual perception can be achieved through hearing needs to be reinforced or enhanced by techniques that lie beyond mere visual-to-audio mapping (e.g. computer vision, image processing). Findings – Experiments reported in this paper reveal that the See ColOr is learnable and functional, and provides ea...

/pdf/see-color-an-extended-sensory-substitution-device-for-the-3crmjbrp6e.pdf

See ColOr: an extended sensory substitution device for the visually impaired

A brain-computer interface (BCI) allows a user to communicate with a computer by means only of the electrical information provided by the brain (EEG), without activation of the motor system. We propose a model for average trial protocols based BCI's. Using this model, we show that, under the hypothesis that the user can emit the same mental state several times, (a) such average trial protocols allow to increase the bit rate B, and (b) the optimal classification speed V leading to the best B can be predicted.

/pdf/information-theoretic-bit-rate-optimization-for-average-578vkre7zp.pdf

Information theoretic bit-rate optimization for average trial protocol Brain-Computer Interfaces

Emotions are often perceived by humans through a series of multimodal cues, such as verbal expressions, facial expressions and gestures. In order to recognise emotions automatically, reliable emotional labels are required to learn a mapping from human expressions to corresponding emotions. Dimensional emotion models have become popular and have been widely applied for annotating emotions continuously in the time domain. However, the statistical relationship between emotional dimensions is rarely studied. This paper provides a solution to automatic emotion recognition for the Audio/Visual Emotion Challenge (AVEC) 2018. The objective is to find a robust way to detect emotions using more reliable emotion annotations in the valence and arousal dimensions. The two main contributions of this paper are: 1) the proposal of a new approach capable of generating more dependable emotional ratings for both arousal and valence from multiple annotators by extracting consistent annotation features; 2) the exploration of the valence and arousal distribution using outlier detection methods, which shows a specific oblique elliptic shape. With the learned distribution, we are able to detect the prediction outliers based on their local density deviations and correct them towards the learned distribution. The proposed method performance is evaluated on the RECOLA database containing audio, video and physiological recordings. Our results show that a moving average filter is sufficient to remove the incidental errors in annotations. The unsupervised dimensionality reduction approaches could be used to determine a gold standard annotations from multiple annotations. Compared with the baseline model of AVEC 2018, our approach improved the arousal and valence prediction of concordance correlation coefficient significantly to respectively 0.821 and 0.589.

/pdf/towards-a-better-gold-standard-denoising-and-modelling-519mhfspw6.pdf

Towards a Better Gold Standard: Denoising and Modelling Continuous Emotion Annotations Based on Feature Agglomeration and Outlier Regularisation

In this paper we consider the problem of capacity analysis in the framework of information-theoretic model of data hiding. Capacity is determined by the stochastic model of the host image, by the distortion constraints and by the side information about watermarking channel state available at the encoder and at the decoder. We emphasize the importance of proper modeling of image statistics and outline the possible decrease in the expected fundamental capacity limits, if there is a mismatch between the stochastic image model used in the hider/attacker optimization game and the actual model used by the attacker. To obtain a realistic estimation of pssible embedding rates we propose a novel stochastic non-stationary image model that is based on geometrical priors. This model outperforms the previously analyzed EQ and spike models in reference application such as denoising. Finally, we demonstrate how the proposed model influences the estimation of capacity for real images. We extend our model to different transform domains that include orthogonal, biorthogonal and overcomplete data representations.

https://archive-ouverte.unige.ch/unige:47963/ATTACHMENT01

Thierry Pun

Papers

Information-transfer rate modeling of EEG-based synchronized brain-computer interfaces

See ColOr: an extended sensory substitution device for the visually impaired

Information theoretic bit-rate optimization for average trial protocol Brain-Computer Interfaces

Towards a Better Gold Standard: Denoising and Modelling Continuous Emotion Annotations Based on Feature Agglomeration and Outlier Regularisation

Data hiding capacity analysis for real images based on stochastic nonstationary geometrical models