262,250 research outputs found
Detection of human activities in natural environments based on their acoustic emissions
No full textPublication in the conference proceedings of EUSIPCO, Bucharest, Romania, 201
Canonical correlation analysis for classifying baby crying sound events
No full textThe present paper concentrates on classifying the state of an infant based on the content of the associated vocalizations. The specific problem belongs to the paralinguistic audio signal processing domain while i has not been studied extensively. Since the specific problem is connected to atypical vocalic expressions, i.e. the sounds produced by the infant are made under stressful conditions we employed the Teager Energy Operator combined with the Mel Frequency Cepstral Coefficients. However these two sets may include overlapping information and therefore provide a redundant feature set if used concurrently. In order to capture the most discriminative information with the minimum number of dimensions we applied canonical correlation analysis on the extracted feature sets. Canonical correlation analysis searches for the correlations between two sets of multidimensional variables and projects them onto a lower-dimensional space in which they are maximally correlated. Subsequently we model the feature space using the Support Vector Machine with a linear kernel. We thoroughly evaluated the proposed methodology on a real-world dataset and present the results in the confusion matrix form. The dataset includes the following five different states: a) hungry, b) uncomfortable (need change), c) need to burp, d) in pain, e) need to sleep. Ultimately the goal of the system is to become an automatic and non-invasive framework for monitoring infants as well as helping pediatricians to better understand their status
A Concept Drift-Aware DAG-Based Classification Scheme for Acoustic Monitoring of Farms
No full textIntelligent farming as part of the green revolution is advancing the world of agriculture in such a way that farms become dynamic, with the overall scope being the optimization of animal production in an eco-friendly way. In this direction, this study proposes exploiting the acoustic modality for farm monitoring. Such information could be used in a stand-alone or complimentary mode to monitor the farm constantly at a great level of detail. To this end, the authors designed a scheme classifying the vocalizations produced by farm animals. More precisely, a directed acyclic graph was proposed, where each node carries out a binary classification task using hidden Markov models. The topological ordering follows a criterion derived from the Kullback-Leibler divergence. In addition, a transfer learning-based module for handling concept drifts was proposed. During the experimental phase, the authors employed a publicly available dataset including vocalizations of seven animals typically encountered in farms, where promising recognition rates were reported
Few-shot learning for modeling cyber physical systems in non-stationary environments
No full textThis paper proposes a modeling scheme for cyber physical systems operating in non-stationary, small data environments. Unlike the traditional modeling logic, we introduce the few-shot learning paradigm, the operation of which is based on quantifying both similarities and dissimilarities. As such, we designed a suitable change detection mechanism able to reveal previously unknown operational states, which are incorporated in the dictionary online. We elaborate on spectrograms extracted from high-resolution ultrasound depth sensor timeseries, while the backbone of the proposed method is a Siamese Neural Network. The experimental scenario considers data representing liquid containers for fuel/water when the following five operational states are present: normal, accident, breakdown, sabotage, and cyber-attack. Thorough experiments were carried out assessing every aspect of the present framework and demonstrating its efficacy even when very few samples per class are available. In addition, we propose a probabilistic data selection scheme facilitating one-shot learning. Last but not least, responding to the wide requirement for interpretable AI, we explain the obtained predictions by examining the layer-wise activation maps
Transfer Learning for Improved Audio-Based Human Activity Recognition
No full textHuman activities are accompanied by characteristic sound events, the processing of which might provide valuable information for automated human activity recognition. This paper presents a novel approach addressing the case where one or more human activities are associated with limited audio data, resulting in a potentially highly imbalanced dataset. Data augmentation is based on transfer learning; more specifically, the proposed method: (a) identifies the classes which are statistically close to the ones associated with limited data; (b) learns a multiple input, multiple output transformation; and (c) transforms the data of the closest classes so that it can be used for modeling the ones associated with limited data. Furthermore, the proposed framework includes a feature set extracted out of signal representations of diverse domains, i.e., temporal, spectral, and wavelet. Extensive experiments demonstrate the relevance of the proposed data augmentation approach under a variety of generative recognition schemes
A Statistical Inference Framework for Understanding Music-Related Brain Activity
No full textFollowing the success in Music Information Retrieval (MIR), research is now steering towards understanding the relationship existing between brain activity and the music stimuli causing it. To this end, a new MIR topic has emerged, namely Music Imagery Information Retrieval, with its main scope being to bridge the gap existing between music stimuli and its respective brain activity. In this paper, the encephalographic modality was chosen to capture brain activity as it is more widely available since of-the-shelf devices recording such responses are already affordable unlike more expensive brain imaging techniques. After defining three tasks assessing different aspects of the specific problem (stimuli identification, group and meter classification), we present a common method to address them, which explores the temporal evolution of the acquired signals. In more detail, we rely on the parameters of linear time-invariant models extracted out of electroencephalographic responses to heterogeneous music stimuli. Subsequently, the probability density function of such parameters is estimated by hidden Markov models taking into account their succession in time. We report encouraging classification rates in the above-mentioned tasks suggesting the existence of an underlying relationship between music stimuli and their electroencephalographic responses
Automatic acoustic identification of respiratory diseases
No full textSeveral disease affecting the human respiratory system, such as asthma, pneumonia, etc. are associated with distinctive sounds. Here, we propose a methodology towards their automatic identification by means of signal processing and pattern recognition algorithms. We designed a suitable feature set based on wavelet packet analysis characterizing data coming from diverse classes of respiratory sounds following the logic of the challenge organised within the International Conference on Biomedical Health Informatics in 2017. The patterns revealed by the feature extraction stage are modelled by hidden Markov models. Automatic identification is carried out via a directed acyclic graph (DAG) scheme limiting the problem space while based on decisions made by the available HMMs. Thorough experiments following a well-established protocol demonstrate the efficacy of the proposed solution. Indeed, such a DAG-based structure outperforms the current state of the art including deep networks based on convolutional kernels. Importantly, the presented solution offers a high level of explainability as one is able to investigate the effective DAG path and understand both correct and incorrect predictions
Acoustic detection of unknown bird species and individuals
No full textComputational bioacoustics is a relatively young research area, yet it has increasingly received attention over the last decade because it can be used in a wide range of applications in a cost-effective manner. This work focuses on the problem of detecting the novel bird calls and songs associated with various species and individual birds. To this end, variational autoencoders, consisting of deep encoding-decoding networks, are employed. The encoder encompasses a series of convolutional layers leading to a smooth high-level abstraction of log-Mel spectrograms that characterise bird vocalisations. The decoder operates on this latent representation to generate each respective original observation. Novel species/individual detection is carried out by monitoring and thresholding the expected reconstruction probability. We thoroughly evaluate the proposed method on two different data sets, including the vocalisations of 11 North American bird species and 16 Athene noctua individuals
On predicting the unpleasantness level of a sound event
No full textThis work presents a novel framework for the automatic assessment of the unpleasantness caused by audio events to a human listener which is a relatively new research problem. Melfrequency cepstral coefficients and temporal modulation parameters were employed to characterize 75 sound stimuli varying from animal calls to baby cries. The final assessment is made by means of a clustering scheme realized by Gaussian mixture models. The proposed framework leads to the best performance in terms of mean squared error and correlation between predicted and measured unpleasantness levels reported so far in the literature
Classification of Sounds Indicative of Respiratory Diseases
No full textThis work presents a system achieving classification of respiratory sounds directly related to various diseases of the human respiratory system, such as asthma, COPD, and pneumonia. We designed a feature set based on wavelet packet analysis characterizing data coming from four sound classes, i.e. crack, wheeze, normal, crack+wheeze. Subsequently, the captured temporal patterns are learned by hidden Markov models (HMMs). Finally, classification is achieved via a directed acyclic graph scheme limiting the problem space while based on decisions made by the available HMMs. Thorough experiments following a well-established protocol demonstrate the efficacy of the proposed solution
- …
