Raviv Raich

Acoustic classification of multiple simultaneous bird species: A multi-instance multi-label approach

Although field-collected recordings typically contain multiple simultaneously vocalizing birds of different species, acoustic species classification in this setting has received little study so far. This work formulates the problem of classifying the set of species present in an audio recording using the multi-instance multi-label (MIML) framework for machine learning, and proposes a MIML bag generator for audio, i.e., an algorithm which transforms an input audio signal into a bag-of-instances representation suitable for use with MIML classifiers. The proposed representation uses a 2D time-frequency segmentation of the audio signal, which can separate bird sounds that overlap in time. Experiments using audio data containing 13 species collected with unattended omnidirectional microphones in the H. J. Andrews Experimental Forest demonstrate that the proposed methods achieve high accuracy (96.1% true positives/negatives). Automated detection of bird species occurrence using MIML has many potential applications, particularly in long-term monitoring of remote sites, species distribution modeling, and conservation planning.

A Hammerstein predistortion linearization design based on the indirect learning architecture

Power amplifiers (PAs) are inherently nonlinear devices and are used in virtually all communications systems. Digital baseband predistortion is a highly cost effective way to linearize the PAs, but most existing architectures assume that the PA has a memoryless nonlinearity. For wider bandwidth applications such as WCDMA, PA memory effects can no longer be ignored, and memoryless predistortion has limited effectiveness. In this paper, we model the PA as a Wiener system and construct a Hammerstein predistorter, obtained using an indirect learning architecture. Linearization performance is demonstrated on a 3-carrier UMTS signal.

Orthogonal polynomials for complex Gaussian processes

Power amplifiers are the major source of nonlinearity in communications systems. Such nonlinearity causes spectral regrowth as well as in-band distortion, which leads to adjacent channel interference and increased bit error rate. Polynomials are often used to model the nonlinear power amplifier or its predistortion linearizer. In this paper, we present a novel set of orthogonal polynomials for baseband Gaussian input to replace the conventional polynomials and show how they alleviate the numerical instability problem associated with the conventional polynomials. The orthogonal polynomials also provide an intuitive means of spectral regrowth analysis.

Spectral analysis of polynomial nonlinearity with applications to RF power amplifiers

The majority of the nonlinearity in a communication system is attributed to the power amplifier (PA) present at the final stage of the transmitter chain. In this paper, we consider Gaussian distributed input signals (such as OFDM), and PAs that can be modeled by memoryless or memory polynomials. We derive closed-form expressions of the PA output power spectral density, for an arbitrary nonlinear order, based on the so-called Leonov-Shiryaev formula. We then apply these results to answer practical questions such as the contribution of AM/PM conversion to spectral regrowth and the relationship between memory effects and spectral asymmetry.

On the baseband representation of a bandpass nonlinearity

Modeling, analysis, and compensation of nonlinearities in the transmitter (the power amplifier, in particular) has attracted a lot of attention recently. Given the same bandpass polynomial or Volterra representation of the nonlinear system, two different baseband formulations have appeared in the literature. The purpose of this correspondence is to examine the discrepancy between the two and to affirm that proper conjugation must be applied in the baseband representation of bandpass nonlinearities. A predistortion linearization test-bed experiment is conducted to illustrate the concepts.

On Local Intrinsic Dimension Estimation and Its Applications

In this paper, we present multiple novel applications for local intrinsic dimension estimation. There has been much work done on estimating the global dimension of a data set, typically for the purposes of dimensionality reduction. We show that by estimating dimension locally, we are able to extend the uses of dimension estimation to many applications, which are not possible with global dimension estimation. Additionally, we show that local dimension estimation can be used to obtain a better global dimension estimate, alleviating the negative bias that is common to all known dimension estimation algorithms. We illustrate local dimension estimations uses towards additional applications, such as learning on statistical manifolds, network anomaly detection, clustering, and image segmentation.

FINE: Fisher Information Nonparametric Embedding

We consider the problems of clustering, classification, and visualization of high-dimensional data when no straightforward Euclidean representation exists. In this paper, we propose using the properties of information geometry and statistical manifolds in order to define similarities between data sets using the Fisher information distance. We will show that this metric can be approximated using entirely nonparametric methods, as the parameterization and geometry of the manifold is generally unknown. Furthermore, by using multidimensional scaling methods, we are able to reconstruct the statistical manifold in a low-dimensional Euclidean space; enabling effective learning on the data. As a whole, we refer to our framework as Fisher information nonparametric embedding (FINE) and illustrate its uses on practical problems, including a biomedical application and document classification.

On the modeling of memory nonlinear effects of power amplifiers for communication applications

Understanding power amplifier (PA) nonlinearity is a first step towards linearization efforts. We first explore the passband and baseband PA input/output relationships and show that they manifest differently when the PA exhibits long-term, short-term, or no memory effects. We then explain the various memory effects in the context of AM/AM and AM/PM responses. The so-called quasi-memoryless case is especially clarified. Four particular nonlinear models with memory are further investigated.

Time-frequency segmentation of bird song in noisy acoustic environments

Recent work in machine learning considers the problem of identifying bird species from an audio recording. Most methods require segmentation to isolate each syllable of bird call in input audio. Energy-based time-domain segmentation has been successfully applied to low-noise, single-bird recordings. However, audio from automated field recorders contains too much noise for such methods, so a more robust segmentation method is required. We propose a supervised time-frequency audio segmentation method using a Random Forest classifier, to extract syllables of bird call from a noisy signal. When applied to a test data set of 625 field-collected audio segments, our method isolates 93.6% of the acoustic energy of bird song with a false positive rate of 8.6%, outperforming energy thresholding.

Audio Classification of Bird Species: A Statistical Manifold Approach

Our goal is to automatically identify which species of bird is present in an audio recording using supervised learning. Devising effective algorithms for bird species classification is a preliminary step toward extracting useful ecological data from recordings collected in the field. We propose a probabilistic model for audio features within a short interval of time, then derive its Bayes risk-minimizing classifier, and show that it is closely approximated by a nearest-neighbor classifier using Kullback-Leibler divergence to compare histograms of features. We note that feature histograms can be viewed as points on a statistical manifold, and KL divergence approximates geodesic distances defined by the Fisher information metric on such manifolds. Motivated by this fact, we propose the use of another approximation to the Fisher information metric, namely the Hellinger metric. The proposed classifiers achieve over 90% accuracy on a data set containing six species of bird, and outperform support vector machines.