The problem of learning fuzzy rule-bases is analyzed from the perspective of finding a favorable balance between the accuracy of the system, the speed required to learn the rules, and finally, the interpretability of the rule-bases obtained. Therefore, we introduce a complete design procedure to learn and then optimize the system rule-base, called the Precise and Fast Fuzzy Modeling approach.Under this paradigm, fuzzy rules are generated from numerical data using a parameterizable greedybased learning method called Selection-Reduction, whose accuracy-speed efficiency is confirmed through empirical results and comparisons with reference methods. Qualitative justification for this method is

Humans express and perceive emotions in a multimodal manner. The multimodal information is intrinsically fused by the human sensory system in a complex manner. Emulating a temporal desynchronisation between modalities, in this paper, we design an end-to-end neural network architecture, called TA-AVN, that aggregates temporal audio and video information in an asynchronous setting in order to determine the emotional state of a subject. The feature descriptors for audio and video representations are extracted using simple Convolutional Neural Networks (CNNs), leading to real-time processing. Undoubtedly, collecting annotated training data remains an important challenge when training emotion recognition systems, both in terms of effort and expertise required. The proposed approach solves this problem by providing a natural augmentation technique that allows achieving a high accuracy rate even when the amount of annotated training data is limited. The framework is tested on three challenging multimodal reference datasets for the emotion recognition task, namely the benchmark datasets CREMA-D and RAVDESS, and one dataset from the FG2020's challenge related to emotion recognition. The results prove the effectiveness of our approach and our end-to-end framework achieves state-of-the-art results on the CREMA-D and RAVDESS datasets.INDEX TERMS Emotion recognition, multimodal data, audiovisual information, augmentation techniques, convolutional neural network, real-time processing.

International audienceWith the increased popularity of touch-sensitive surfaces, much attention has been drawn to their security-related issues, as they currently rely only on the visual sense for feedback. To improve operability, vibrotactile signals may be delivered to the finger on screen interaction. The way vibrotactile signals affect human perception is examined via three measured variables, related to their energy, velocity, and spectral complexity, and which are analytically defined in this paper. It is shown that these variables accurately account for the psychophysical properties of the tactile sense. Based on this, a psychophysical fuzzy rule-based model of vibrotactile perception is introduced to forecast the comfort values of the vibrational signals provided by an automobile haptic screen. Using an efficient rule-based generation method, a Mamdani fuzzy inference system is proposed; it achieves a mean error rate of 14% for the train set and 17% for the test set, while correctly classifying most of the signals within a reasonable tolerance, related to human evaluation imprecision. The system also produces a comprehensible linguistic rule structure, which allows behavioral patterns to be detected

International audienceRule base generation from numerical data has been a dynamic research topic within the fuzzy community in the last decades, and several well-established methods have been proposed. While some authors presented simple, empirical approaches, but which generally show high error rates, others turned to complex heuristic techniques to improve accuracy. In this paper, an extension of the classical Wang-Mendel method is proposed. While keeping a linear complexity, the new method achieves performances close to those of more complex methods based on cooperative rules (COR). Results on synthetic data show the potential of the proposed method as a complexity-accuracy trade-off

No abstract