Learning to disentangle and represent factors of variation in data is an important problem in artificial intelligence. While many advances have been made to learn these representations, it is still unclear how to quantify disentanglement. While several metrics exist, little is known on their implicit assumptions, what they truly measure, and their limits. In consequence, it is difficult to interpret results when comparing different representations. In this work, we survey supervised disentanglement metrics and thoroughly analyze them. We propose a new taxonomy in which all metrics fall into one of the three families: intervention-based, predictor-based, and information-based. We conduct extensive experiments in which we isolate properties of disentangled representations, allowing stratified comparison along several axes. From our experiment results and analysis, we provide insights on relations between disentangled representation properties. Finally, we share guidelines on how to measure disentanglement.

Dynamic and nonlinear systems are emerging as potential candidates for random bit generation. In this context, chaotic systems, which are both dynamic and stochastic, are particularly suitable. This article introduces a new continuous chaotic system along with its corresponding implementation, which targets FPGA (fpga). This chaotic system has five dimensions, which exhibit complex chaotic dynamics, thus enabling the utilization of chaotic signals in cryptography. A mathematical analysis is presented to demonstrate the dynamic characteristics of the proposed hyperchaotic system. A novel digital implementation of the proposed system is presented. Moreover, a data scrambling circuit is implemented to eliminate the bias effect and increase the randomness of the bitstream generated from the chaotic signals. We show that the proposed random bit generator has high randomness. The generated bits successfully pass well-known statistical randomness test-suites, i.e., NIST SP800-22, Diehard, and TestU01. The ready-to-use random bit generator is deployed on a Xilinx Zynq-7000 SoC ZC702 Evaluation Kit. Experimental results show that the proposed random bit generator can achieve a maximum throughput of 6.78 Gb/s, which is over 3.6 times greater than state-of-the-art designs, while requiring under 4% of the resources available on the targeted FPGA.

Capacitive electrocardiogram (cECG) systems are increasingly used for the monitoring of cardiac activity. They can operate within the presence of a small layer of air, hair or cloth and do not require a qualified technician. They can be integrated into wearables, clothing or objects of daily life, such as beds or chairs. While they offer many advantages over conventional electrocardiogram systems (ECG) that rely on wet electrodes, they are more prone to be affected by motion artifacts (MAs). These effects, which are due to the relative movement of the electrode in relation to the skin, are several orders of magnitude higher than ECG signal amplitudes, they occur in frequencies that might overlap with the ECG signal, and they may saturate the electronics in the most severe cases. In this paper, we provide a detailed description of MA mechanisms that translate into capacitance variations due to electrode-skin geometric changes or into triboelectric effects due to electrostatic charge redistribution. A state-of-the-art overview of the different approaches based on materials and construction, analog circuits and digital signal processing is provided as well as the trade-offs to be made using these techniques, to mitigate MAs efficiently.

In this work, we investigate a scenario where both multistatic radar and point-to-point communication systems are present with partial or total spectrum sharing constraint. We propose a new adaptive radar receivers placement mechanism that jointly maximizes the signal to interference plus noise ratio (SINR) of each communication transmitter-radar receiver channel while minimizing the geometric dilution of precision (GDOP). The proposed joint approach performs well in presence of communication interference on the radar side. In fact, the proposed approach helps to increase the capability to properly demodulate communication data at each radar receiver resulting in less radar measurement errors due to communication interference, while enhancing the target positioning accuracy.

Several methods have recently been proposed to analyze speech and automatically infer the personality of the speaker. These methods often rely on prosodic and other hand crafted speech processing features extracted with off-the-shelf toolboxes. To achieve high accuracy, numerous features are typically extracted using complex and highly parameterized algorithms. In this paper, a new method based on feature learning and spectrogram analysis is proposed to simplify the feature extraction process while maintaining a high level of accuracy. The proposed method learns a dictionary of discriminant features from patches extracted in the spectrogram representations of training speech segments. Each speech segment is then encoded using the dictionary, and the resulting feature set is used to perform classification of personality traits. Experiments indicate that the proposed method achieves state-of-the-art results with an important reduction in complexity when compared to the most recent reference methods. The number of features, and difficulties linked to the feature extraction process are greatly reduced as only one type of descriptors is used, for which the 7 parameters can be tuned automatically. In contrast, the simplest reference method uses 4 types of descriptors to which 6 functionals are applied, resulting in over 20 parameters to be tuned.

A growing number of applications, e.g., video surveillance and medical image analysis, require training recognition systems from large amounts of weakly annotated data, while some targeted interactions with a domain expert are allowed to improve the training process. In such cases, active learning (AL) can reduce labeling costs for training a classifier by querying the expert to provide the labels of most informative instances. This paper focuses on AL methods for instance classification problems in multiple instance learning (MIL), where data are arranged into sets, called bags, which are weakly labeled. Most AL methods focus on single-instance learning problems. These methods are not suitable for MIL problems because they cannot account for the bag structure of data. In this paper, new methods for bag-level aggregation of instance informativeness are proposed for multiple instance AL (MIAL). The aggregated informativeness method identifies the most informative instances based on classifier uncertainty and queries bags incorporating the most information. The other proposed method, called cluster-based aggregative sampling, clusters data hierarchically in the instance space. The informativeness of instances is assessed by considering bag labels, inferred instance labels, and the proportion of labels that remain to be discovered in clusters. Both proposed methods significantly outperform reference methods in extensive experiments using benchmark data from several application domains. Results indicate that using an appropriate strategy to address MIAL problems yields a significant reduction in the number of queries needed to achieve the same level of performance as single-instance AL methods.