Canonical correlation analysis as a feature extraction method to classify active sonar targets with shallow neural networks.

Sonar target recognition remains an active area of research due to the complex entanglement of features from various acoustic scatterers, background clutter, and distortion by waveguide propagation effects. An equally challenging issue is due to different acoustic echoes returned from the target (including different target elements) itself. This work investigates the sonar target classification problem from a statistical perspective and aims to extract salient target feature vectors. Specifically, a multivariate statistical method is employed, canonical correlation analysis (CCA), as a feature extraction technique prior to multi-class classification of active sonar field data. The intuition behind using CCA is that persistent features slowly morph over time due to the changing aspect angles and platform positions and can be represented by maximally correlated projections of consecutive pings. CCA is applied using a sliding window, and the projections are used as feature vectors to train a neural network classifier. The smallest increase in classification accuracy when comparing the projection feature vectors to unprocessed feature vectors was 10%. The largest increase was 34%. The results are further examined through the use of confusion matrices and layer-wise relevance propagation, which distributes the trained networks output score to the input layer.

Signal Processing Methods to Interpret Polychlorinated Biphenyls in Airborne Samples.

The main contribution of this interdisciplinary work is a robust computational framework to autonomously discover and quantify previously unknown associations between well-known (target) and potentially unknown (non-target) toxic industrial air pollutants. In this work, the variability of polychlorinated biphenyl (PCB) data is evaluated using a combination of statistical, signal processing, and graph-based informatics techniques to interpret the raw instrument signal from gas chromatography-mass spectrometry (GC/MS/MS) data sets. Specifically, minimum mean-squared techniques from the adaptive signal processing literature are extended to detect and separate coeluted (overlapped) peaks in the raw instrument signal. A graph-based visualization is provided which bridges two complementary approaches to quantitative pollution studies: (i) peak-cognizant target analysis (limits data analysis to few well-known compounds) and (ii) chemometric analysis (statistical large-scale data analysis) that is agnostic of specific compounds. Further, peak fitting techniques based on L2 error minimization are employed to autonomously calculate the amount of each PCB present with a normalized mean square error of -18.4851 dB. Graph-based visualization of associations between known and unknown compounds are developed through principal component analysis and both fuzzy c-means (FCM) and k-means clustering techniques are implemented and compared. The efficiency of these methods are compared using 150 air samples analyzed for individual PCBs with GC/MS/MS against traditional target-only techniques that perform analysis across only the known (target) PCBs. Parameter optimization techniques are employed to evaluate the relative contribution of PCB signals against ten potential source signals representing legacy signatures from historical manufacture of Aroclors and modern sources of PCBs produced as by products of pigment and polymer manufacturing. Aroclors 1232, 1254, 1016, and 1221 as well as non-Aroclor 3, 3', dichlorobiphenyl (PCB 11) were found in many of the samples as unique source signals that describe PCB mixtures in air samples collected from Chicago, IL.

Sonar target representation using two-dimensional Gabor wavelet features.

This paper introduces a feature extraction technique that identifies highly informative features from sonar magnitude spectra for automated target classification. The approach involves creating feature representations through convolution of a two-dimensional Gabor wavelet and acoustic color magnitudes to capture elastic waves. This feature representation contains extracted localized features in the form of Gabor stripes, which are representative of unique targets and are invariant of target aspect angle. Further processing removes non-informative features through a threshold-based culling. This paper presents an approach that begins connecting model-based domain knowledge with machine learning techniques to allow interpretation of the extracted features while simultaneously enabling robust target classification. The relative performance of three supervised machine learning classifiers, specifically a support vector machine, random forest, and feed-forward neural network are used to quantitatively demonstrate the representations' informationally rich extracted features. Classifiers are trained and tested with acoustic color spectrograms and features extracted using the algorithm, interpreted as stripes, from two public domain field datasets. An increase in classification performance is generally seen, with the largest being a 47% increase from the random forest tree trained on the 1-31 kHz PondEx10 data, suggesting relatively small datasets can achieve high classification accuracy if model-cognizant feature extraction is utilized.