Source Separation Research Articles

Optimizing municipal solid waste management in industrial ports: A case study of source separation implementation

ABSTRACT With the acceleration of industrialization and population growth, there has been a notable surge in municipal solid waste (MSW) production globally. This necessitates the development of efficient waste management strategies that align with sustainable economic, environmental, and social goals. Implementing sorting and source separation policies serves as foundational steps toward optimizing material flow and resource utilization. In industrial areas, waste generation around a primary source may improve policy implementation and participation. Given the unique strategic context of port environments, Bahregan Port in Iran was selected as a case study to evaluate the proposed methodology. Three integrated scenarios were formulated and prioritized using the ELECTRE-I model: currently practiced collection method, and segregated collection followed by source separation of MSW into two or three groups. The analysis identified the third scenario as the most effective approach. Remarkably, when evaluating the results under optimistic and pessimistic conditions, the ranking of preferences remained consistent across scenarios in the optimal state. Under pessimistic conditions, all scenarios received identical scores, indicating significant uncertainty in criteria weights. Furthermore, sensitivity analysis revealed that criteria such as the raw material sources protection, emissions in the waste separation sector, and greenhouse gas emissions were highly influential. Interestingly, criteria with the highest weighting did not always dictate the final scenario selection. External factors, such as expert opinions on specific criteria, played a pivotal role in scenario prioritization. This research serves as a foundational study for source separation initiatives, offering practical and technically viable strategies to enhance environmental sustainability within industrial settings. Implications Statement Rising industrialization and population growth have led to increased municipal solid waste (MSW) production worldwide. To address this, efficient waste management strategies are essential, focusing on sustainability and resource optimization through sorting and source separation policies. By prioritizing the scenario of segregated collection followed by source separation – industrial areas can optimize material flow and resource utilization. The study highlights the influence of criteria like raw material protection, emissions, and greenhouse gases, making this research foundational for enhancing environmental sustainability in industrial settings. This study provides practical strategies for enhancing environmental sustainability in industrial settings through source separation initiatives.

A Non-Invasive Fetal QRS Complex Detection Method Based on a Multi-Feature Fusion Neural Network

Fetal heart monitoring, as a crucial part of fetal monitoring, can accurately reflect the fetus’s health status in a timely manner. To address the issues of high computational cost, inability to observe fetal heart morphology, and insufficient accuracy associated with the traditional method of calculating the fetal heart rate using a four-channel maternal electrocardiogram (ECG), a method for extracting fetal QRS complexes from a single-channel non-invasive fetal ECG based on a multi-feature fusion neural network is proposed. Firstly, a signal entropy data quality detection algorithm based on the blind source separation method is designed to select maternal ECG signals that meet the quality requirements from all channel ECG data, followed by data preprocessing operations such as denoising and normalization on the signals. After being segmented by the sliding window method, the maternal ECG signals are calculated as data in four modes: time domain, frequency domain, time–frequency domain, and data eigenvalues. Finally, the deep neural network using three multi-feature fusion strategies—feature-level fusion, decision-level fusion, and model-level fusion—achieves the effect of quickly identifying fetal QRS complexes. Among the proposed networks, the one with the best performance has an accuracy of 95.85% and sensitivity of 97%.

A positive semi definite tensor factorization method for separation of non stationary noise sources

In the aircraft industry, noise mitigation has emerged as an increasingly pressing issue, underscoring the critical importance of advancing our understanding of noise origins within turbofan engines. This paper presents the application of Positive Semi Definite Tensor Factorization (PSDTF), a potential method for the analysis of engine static tests conducted with far-field microphone arrays. By extending the capabilities of Non-negative Matrix Factorization (NMF), PSDTF offers an effective algorithm for source separation. Leveraging on cross spectral matrices to harness phase information across microphones, this approach aims at separating the contributions of several noise sources, avoiding the need for a precise acoustical model (sound propagation, source directivity, etc.). Experimental findings on a controlled experiment demonstrate the superiority of PSDTF over conventional NMF variants in achieving higher-quality source separation.

Rotating and stationary sound sources separation method based on MCB

Traditional beamforming methods do not effectively separate the mixed field of sound sources for different rotational speeds. The coexistence of these sources often manifests out-of-focus sources (for instance, stationary sources appearing with a rotational focus), thereby blurring the beamforming outcomes. Most existing algorithms for segregating rotational different speed sources employ a virtual rotating array utilizing ring arrays. However, due to the constant microphone spacing, the acoustic imaging of the ring array suffers from the interference of high-level side lobes. In this work, a hybrid deconvolution method based on modal composition beamforming is used to explore the ability of a multi-arm spiral array to separate different speed sources. Considering the even energy distribution from the out-of-focus source along the circular trajectory, the beamforming result of out-of-focus sources exhibits a weaker amplitude than the source amplitude. Therefore, out-of-focus sources can be considered noise. First, the point spread function is derived based on the modal transfer function. Then, a system of linear equations is constructed based on an incomplete cross-point spread function matrix. Finally, the source separation is solved through sparse-constrained deconvolution. Simulation and experimental results demonstrate that this method can accurately locate and separate different rotational speed sound sources.

Deep learning-enhanced blind separation of incoherent and spatially disjoint sound sources

Separating incoherent sound sources within complex acoustical fields presents a significant challenge in acoustic imaging. Existing methods, such as Principal Component Analysis (PCA) applied to the Cross-Spectral Matrix (CSM), yield 'virtual' sources based on statistical orthogonality. However, this approach often fails to accurately identify distinct physical sources, primarily due to its reliance on statistical orthogonality solely. A state-of-the-art method involves computing a rotation matrix to enforce criteria such as least spatial entropy, or spatial orthogonality among sources, a process that, while effective, significantly increases computational complexity and time. This work introduces a hybrid approach combining PCA and deep learning for predicting spatially disjoint source maps from virtual sources. By simulating sound sources in random quantities and locations, we train a neural network tailored to this task. We address the order mismatch between PCA-derived virtual sources and pre-simulated labels by framing source separation as a set prediction problem, utilizing the Hungarian loss for efficient mismatch resolution. This method simplifies the separation process, offering faster post-training computations and eliminating the need for complex optimizations. Validation in simulated environments and real-world datasets has shown the model's effectiveness in source separation for acoustic imaging, indicating the potential of integrating deep learning with existing methods.

Dynamic localization of unwanted events with machine learning-based method

In environmental acoustics measurements it is necessary to separate the contribution of the investigated source from the residual noise, by means of ambient noise measurements. Source separation is a complex task, especially for wind turbine noise, where long-term measurement campaigns are required. Although data analysis procedures that provide for this task are already in use, all of them require a preliminary step for anomalous events removal, i.e. events which are not characteristic of standard residual noise. This step can be complex and very time-expensive depending on the duration of the measurements and the measurement site. Depending on the specific site and source under evaluation, the variety of sounds that can be considered as unwanted for compliance with environmental acoustic regulations is heterogeneous, ranging from road vehicles, trains and planes to human and animal sounds and more. Nowadays, it is possible to extrapolate from an audio source the features that describe its semantics using machine learning. The approach presented in this paper aims to localize in time a specific event within a measurement containing several acoustic events, in order to label it and then proceed to remove the spurious events from the measurement.

Development of an experimental technique for measuring polar diagrams of adaptive directivity implemented in hearing aids

Nowadays, hearing aids have efficient signal processing options to improve speech intelligibility in noise. This work concerns the characterization of the microphone systems directivity available in hearing aids. In recent decades, "adaptive" directivities have been implemented in hearing aids, thus making it possible to focus on useful sound sources in the hearing aid patient environment. These options use the principles of sound environments detection, acoustic beamforming and noise reduction. Performances of these options can be estimated with subjective methods (as speech audiometry in noise), and objective methods (listening effort, intelligibility indicators as HASPI, SII, STI, etc.). But these methods provide an overall assessment of all the processing involved in speech enhancement and are not designed to characterize the effects of adaptive directivity. We developed an experimental technique based on sound sources separation with phase opposition diffusion (known as "Hagerman" method) to draw polar diagrams of adaptive directivity, for realistic sound environments. We used a sound multi-diffusion system to generate various sound environments with noise and speech sources around an artificial head equipped with hearing aids. The first results show the adaptive directivity effects associated with noise reduction provided by the latest hearing aids generation on everyday hearing scenes.

Diagnosing multiple faults in rotating machinery using empirical mode decomposition and blind source separation methods

It is a challenging task to accurately diagnose multiple faults in a rotating machinery. In this paper, the authors have used empirical mode decomposition (EMD) and blind source separation (BSS) methods for this purpose. A portable rotating machinery setup is made in the laboratory and faults such as bearing inner and outer race faults, angular misalignment are created into it. The vibration data is collected using SPM made LEONOVA DIAMOND(r) FFT analyzer and processed using CONDMASTER RUBY(r) and MATLAB(r) softwares. Firstly, wavelet transform is applied to the signal to denoise it. After this, EMD is implemented to decompose the signal into different intrinsic mode functions (IMFs). After preprocessing these IMFs using the principal component analysis (PCA), BSS is used to extract fault-related information from the IMFs. A comparison of the results obtained from EMD and PCA-BSS is also presented in the paper. Keywords: Rotating machinery, Multiple faults, Vibration data, Wavelet transform, Empirical mode decomposition, Blind source separation, Principal component analysis.

A-209 Preanalytical Phase Errors Constitute the Vast Majority of Errors in Clinical Laboratory Testing

Abstract Background Laboratory errors can occur during the preanalytical, analytic, or post-analytic phase of testing and may pose a considerable threat to patient safety. Studies have estimated that pre-analytical error encapsulates ∼45-68% of laboratory errors and occurs in up to 1.52% of specimens. However, limitations of previously published studies include the use of manual error detection, reliance on manual transcription, small study sizes, and exclusion of common errors. Herein, we aim to identify the types and frequency of errors occurring in each phase of clinical laboratory testing using a large retrospective dataset across a broad range of errors detected in clinical laboratories. Methods This study was performed in the Barnes-Jewish Hospital core laboratory which provides testing for a 1,400-bed tertiary care hospital and included all tests performed for clinical chemistry, special chemistry, toxicology, automated hematology, blood gas testing, coagulation, non-infectious serology, infectious disease serology, urinalysis, and send out testing between January 2022 and May 2023. Errors were compiled from two separate sources. The first source of errors was captured from the laboratory information system (LIS, Cerner). These errors were flagged manually by laboratory staff in real time. The second source of error was a daily report manually curated by a trained medial laboratory scientist as part of routine, laboratory operating procedures and standard workflow. Results There were ∼40,000,000 individual results from ∼11,000,000 specimens during the study period and 87,317 errors observed (∼0.8% of specimens). Of these errors, specimens that were hemolyzed sufficiently to impact result but still reportable were the majority (41,047, 47.2%). When excluding minimal hemolysis, there were 46,270 errors documented. Among these, 44,847 (97.0%) were in the preanalytical phase, 451 (1.0%) were in the analytical phase, and 972 (2.0%) occurred in the postanalytical phase. Subcategorization of preanalytical errors revealed that most of the errors were related to hemolyzed samples requiring results to be suppressed (n=19,701), quantity not sufficient samples (n=8,068), clotted specimens (n=5,840), collection errors (e.g., wrong tube type, n=5,780), not on ice (n=1,369) and IV fluid contamination (n=1,122). In the analytical phase, the causes of error were interfering substances (n=279), QC out of range (n=65), instrument problems (n=58) and reagent issues (n=49). Among the postanalytical errors, 581 were comment errors, 216 were results entry errors, 131 were comments errors with potential impact on patient care, and 44 were dilution errors. Conclusions This study demonstrates that the vast majority of lab errors (∼97.0%) occurring during the preanalytical phase. The high incidence of preanalytical errors, especially errors related to sample integrity implies a need for enhanced tools and training to minimizing errors from specimen collections, maintenance of sample stability, transportation, and processing.

Multistructured Accretion Flow of Sgr A*. I. Examination of a Radiatively Inefficient Accretion Flow Model

The extreme low-luminosity supermassive black hole Sgr A* provides a unique laboratory in which to test models of radiatively inefficient accretion flows (RIAFs). Previous fits to the quiescent Chandra ACIS-S spectrum found that a RIAF model with an equal inflow–outflow balance works well. In this work, we apply the RIAF model to the Chandra HETG-S spectrum obtained through the Chandra X-ray Visionary Program, which displays features suggestive of temperature and velocity structures within the plasma. A comprehensive forward model analysis accounting for the accretion flow geometry and HETG-S instrumental effects is required for a full interpretation of the quiescent Chandra HETG-S spectrum. We present a RIAF model that takes these effects into account. Our fits to the high-resolution grating spectrum indicate an inflow balanced by an outflow (s ∼ 1) alongside a temperature profile that appears shallower than what would be expected from a gravitational potential following 1/r. The data require that the abundance of iron relative to solar is Z Fe < 0.32 Z ⊙ (90% credible interval), much lower than the 2 Z ⊙ metallicity measured in nearby late-type giants. While future missions like NewAthena will provide higher spectral resolution, source separation will continue to be a problem. Leveraging Chandra’s unparalleled spatial resolution, which is not expected to be surpassed for decades, remains essential for detailed investigations of the densely populated Galactic center in X-rays.

Score-matching neural networks for improved multi-band source separation

We present the implementation of a score-matching neural network that represents a data-driven prior for non-parametric galaxy morphologies. The gradients of this prior can be incorporated in the optimization of galaxy models to aid with tasks like deconvolution, inpainting or source separation. We demonstrate this approach with modification of the multi-band modeling framework scarlet that is currently employed as deblending method in the pipelines of the HyperSuprimeCam survey and the Rubin Observatory. The addition of the prior avoids the requirement of non-differentiable constraints, which can lead to convergence failures we discovered in scarlet. We present the architecture and training details of our score-matching neural network and show with simulated Rubin-like observations that using a data-driven prior outperforms the baseline scarlet method in accuracy of total flux and morphology estimates, while maintaining excellent performance for colors. We also demonstrate significant improvements in the robustness to inaccurate initializations. The trained score models used for this analysis are publicly available at https://github.com/SampsonML/galaxygrad.

Carbon footprint impact of waste sorting on the municipal household waste treatment system: A community case study of Hangzhou

Considering the pressing need to reduce carbon emissions, the environmental impacts of waste sorting and its influencing factors remain unverified across many practical scenarios. This study focused on separating perishable and residual waste in a community in Hangzhou, China, examining how waste sorting behavior affects the carbon footprint of subsequent waste treatment systems through a randomized tracking survey and life cycle assessment (LCA) method. The results indicated that the sorted waste treatment system could achieve net carbon emissions ranging from −139.92 to −254.03 kg CO2 eq per day by integrating the anaerobic digestion of sorted perishable waste with the incineration of residual waste. In comparison to the conventional mixed incineration system, which had net carbon emissions between −56.37 and −115.46 kg CO2 eq per day and generated 383 kWh/t of power, the carbon reduction benefits of the sorted system were considerably greater, by a factor of 2.2–2.5. Additionally, the average amount of recovered electrical energy (993 kWh/t) was also higher by a factor of 2.6. Therefore, source separation of waste significantly enhances energy recovery and carbon reduction in household waste treatment systems. In this research system, incineration power generation and oil recovery serve as the primary units for reducing carbon emissions, and the proportion of residual waste generated greatly influences the total carbon emissions of the sorted waste treatment system (p < 0.05). Consequently, increasing the percentage of classified residual waste and optimizing machine efficiency will increase the carbon reduction benefits of waste sorting and treatment systems. These findings provide crucial references and guidance for advancing carbon reduction and waste classification in municipal solid waste (MSW) management systems.

A comprehensive analysis and risk evaluation of microplastics contamination in Australian commercial plant growth substrates: Unveiling the invisible threat

In Australia, quality standards for composts and potting mixes are defined by AS4454-2012 and AS3743-2012. These standards outline key parameters, including physicochemical properties, nutrient content, and plant toxicity. However, they do not address emerging pollutants like microplastics (< 1mm). This study investigates the prevalence and characteristics of MPs in commercial plant growth substrates (PGS), including nineteen potting mixes and five composts, revealing a significant occurrence of MPs, with concentrations ranging from 233 to 7367 particles Kg-1 and an average of 1869 ± 109 particles Kg-1. MPs categorized by shape, size, and color, with fragments (491 ± 34 particles Kg-1), white colour (3700 ± 917 particles Kg-1), and size 500µm being predominant. The polymer composition was diverse, with polyethylene being the most prevalent, followed by polypropylene and others. Polyterpene, Polyalkene, Pentaerythritol, and Propylene glycol were identified in PGS for the first time. The structural equation model showed that physicochemical properties like pH, EC, TOC, and heavy metals influence MPs abundance and characteristics. The Polymer Risk Index and Pollution Load Index indicated varying risk levels among the samples. These findings highlight the need to address MPs contamination in PGS to ensure ecosystem safety and human health. Environmental ImplicationsThe study reveals significant microplastic contamination in Australian commercial PGS, with compost exhibiting higher levels than potting mixes. Application of these contaminated substrates can lead to microplastic release, which can further alter soil properties, disrupt microbial communities, and leach environmental toxins. Current regulations primarily target larger plastic items, neglecting the issue of microplastics (<1mm). The findings underscore the necessity for updated standards and quality controls for commercial PGS. Enhancing public awareness and improving waste sorting are crucial steps to mitigate contamination. Effective source separation is essential for producing high-quality compost and potting mixes. Collaborative efforts among policymakers, industries, and communities are vital for advancing waste management practices and environmental protection.

Noise-robust automated sudden damage detection using blind source separation enhanced by variational mode decomposition and support vector machine based on shapelet transform

Compared to parametric counterparts, non-parametric (aka, model-free) damage detection methods have no requirements of accurate models, with the potential of autonomous monitoring of various complex structures. However, noises, or low signal-to-noise ratio (SNR), are one of the main challenges. This study is aimed at improving blind source separation (BSS)-based damage detection method, one of the most advanced non-parametric methods, in both aspects of noise robustness and autonomous operation. In particular, the measured acceleration responses are processed by variational mode decomposition (VMD) and wavelet transform (WT) in sequential, acting as the input for a BSS model. The BSS is then solved by independent component analysis (ICA), which approves to be more noise-robust compared to the state-of-the-art counterparts. Furthermore, shapelet transform is applied to extract the universal shape-based spike-like feature from the BSS model for training a support vector machine (SVM) model, applicable to different structures; it finally automates the sudden damage detection process and enables online monitoring. The effectiveness of the proposed method is illustrated by a numerical example and an experimental test, and demonstrated by a real-world seismic-excited structure. The results show that both single and multiple sudden damages can be automatically detected with high accuracy. Compared with the existing BSS methods, the proposed BSS method is more capable to detect small damages at relatively low SNR. In addition, the classification accuracy of SVM is also improved when shapelet-based feature is used for training, which reduces the malfunction of automated damage detection as shown by the numerical example. Therefore, the proposed strategy has the potential for rapid condition assessment of structures during rare/extreme events, before engineers are sent for further post-event inspection.

Task-guided attention increases non-linearity of steady-state visually evoked potentials

Objective.Attention is a multifaceted cognitive process, with nonlinear dynamics playing a crucial role. We investigated the involvement of nonlinear processes in top-down visual attention.Approach.The research paradigm employed a contrast-modulated sequence of letters and numerals, encircled by a consistently flickering white square on a black background-a setup that generated steady-state visually evoked potentials. Nonlinear processes are recognized for eliciting and modulating the harmonics of constant frequencies. Using the rhythmic entrainment source separation technique, we examined the fundamental and harmonic frequencies of each stimulus to evaluate the underlying nonlinear dynamics during stimulus processing.Main results.In line with prior research, our findings indicate that the power spectrum density of electroencephalogram responses is influenced by both task presence and stimulus contrast. We discovered that actively searching for a target within a letter stream heightened the amplitude of the fundamental frequency and harmonics related to the background flickering stimulus. While the fundamental frequency amplitude remained unaffected by the stimulus contrast, a lower contrast led to an increase in the second harmonic's amplitude. We assessed the relationship between the contrast response function and the nonlinear-based harmonic responses.Significance.Our findings contribute to a more nuanced understanding of the nonlinear processes impacting top-down visual attention.

Analysis of Cascaded H Bridge Multilevel Inverter for Three-Phase Induction Motor Drive

The last ten years have seen a rise in the usage of multilayer inverters. Because they can produce waveforms with a greater harmonic spectrum and reliable output. Several high voltages and high-power applications are suitable for these new inverters. A multilayer converter not only produce high power ratings, but it also saves time and money and enhances the system performance in terms of harmonics, dv/dt strains, and pressures on the motor bearings. Owing to its flexibility and modularity, the possible topology for a power application is the cascaded H-bridge multilevel inverter (CMI) with separated DC sources. This study investigates a cascade H bridge multilevel inverter, such as a Three, Five, Seven, Nine and Eleven level inverter in order to drive a three phase AC induction motor. A Level shifted Pulse Width Modulation (LSPWM) technique was applied for the CHBMLI. Specific emphasis is given to aspects like modulation and total harmonic distortion that are preferable or necessary for multi-level converters. For both intellectual and scientific considerations, investigating sine triangle carrier pulse width modulation is deemed to be the most auspicious approach. Analysis of the many performance parameters, including load torque, motor speed, and efficiency, an extensive survey is conducted. MATLAB / Simulink is built into the entire system, and simulation results are developed. The purpose of this research is to demonstrate the performance of 9-level and 11-level fed induction motor driving results for Electric Vehicle applications. CHBMLI fed motor drive is better suitable for Induction Motor Drives.

Challenges and opportunities in managing biodegradable plastic waste: A review.

Biodegradable plastics have certain challenges in a waste management perspective. The existing literature reviews fail to provide a consolidated overview of different process steps of biodegradable plastic waste management and to discuss the support provided by the existing legislation for the same. The present review provides a holistic overview of these process steps and a comprehensive relative summary of 13 existing European Union (EU) laws related to waste management and circular economy, and national legislations plus source separation guidelines of 13 countries, to ensure the optimal use of resources in the future. Following were the major findings: (i) numerous types and low volumes of biodegradable plastics pose a challenge to developing cost-effective waste management infrastructure; (ii) biodegradable plastics are promoted as food-waste collection aids, but consumers are often confused about their proper disposal and are prone to greenwashing from manufacturers; (iii) industry-level studies demonstrating mechanical recycling on a full scale are unavailable; (iv) the existing EU legislation dealt with general topics related to biodegradable plastics; however, only the new proposal on plastic packaging waste and the EU policy framework for bioplastics clearly mentioned their disposal and (v) clear disparities were observed between disposal methods suggested by national legislation and available source separation guidelines. Thus, to appropriately manage biodegradable plastic waste, it is necessary to develop waste processing and material utilization infrastructure as well as create consumer awareness. In the end, recommendations were provided for improved biodegradable plastic waste management from the perspective of systemic challenges identified from the literature review.

Occupancy Map Guided Attributes Artifacts Removal for Video-based Point Cloud Compression

Point clouds offer realistic three-dimensional (3-D) representations of objects and scenes at the expense of large data volumes. To represent such data compactly in real-world applications, Video-based Point Cloud Compression (V-PCC) converts their texture into two-dimensional (2-D) attributes and occupancy maps before applying lossy video compression. Unfortunately, the coding artifacts introduced in the decoded attribute maps eventually degrade the quality of the reconstructed point cloud, thereby influencing its immersive experience. This paper proposes a deep learning-based attribute map enhancement method that fully leverages the occupancy map's guidance. The design philosophy is that the cross-modality guidance from occupancy can be leveraged as critical information to enhance the attribute. Therefore, instead of treating attribute and occupancy as two separate sources of signals, occupancy serves as an indispensable auxiliary, such that the proposed framework explicitly provides the model with abundant clues by conducting local feature modification and global dependencies aggregation. In particular, the proposed framework is compatible with existing V-PCC bitstreams and can be feasibly incorporated into the standardized decoder pipeline. Extensive evaluations show the effectiveness of the proposed framework in attribute enhancement, with equivalently 6.0% Bjontegaard Delta-rate (BD-rate) savings obtained.

Unsupervised machine learning with different sampling strategies and topographic factors for distinguishing between landslide source and runout areas to improve landslide inventory production

This study derived 12 topographical and hydrological factors related to landslides from a 10-m digital elevation model. Three unsupervised machine learning algorithms were employed to distinguish between the features of landslide sources and runout areas for Typhoon Morakot. Two sampling strategies were designed in this study. The first strategy involved creating a data set by pairing landslide sources and runout areas on the basis of the size of polygonal areas recorded in the inventory and then exploring the effectiveness of feature separation with k-means++ as the primary method and EM (expectation maximization) and DBSCAN (Density-Based Spatial Clustering of Applications with Noise) as supplementary methods. Because of practical challenges associated with the inability to determine whether remote sensing results correspond to landslide sources or runout areas, the first sampling strategy was used to test the feasibility of the adopted algorithms. In the second sampling strategy, the effectiveness of feature separation was tested by dividing polygon samples into several intervals on the basis of their sizes, thereby verifying the practicability of the adopted algorithms in real-world operations. This study verified that combining unsupervised machine learning algorithms with topographic factors facilitates the effective separation of landslide sources and runout areas, thereby enhancing the quality of landslide inventories and reducing the cost of landslide inventory creation. Moreover, it indicated that meter-scale topographic data yield classification results similar to those obtained from manually created landslide inventories based on centimeter-scale stereo aerial photos. The proposed method effectively balances the cost and quality of landslide inventories.

Semiautomated generation of species-specific training data from large, unlabeled acoustic datasets for deep supervised birdsong isolation.

Bioacoustic monitoring is an effective and minimally invasive method to study wildlife ecology. However, even the state-of-the-art techniques for analyzing birdsongs decrease in accuracy in the presence of extraneous signals such as anthropogenic noise and vocalizations of non-target species. Deep supervised source separation (DSSS) algorithms have been shown to effectively separate mixtures of animal vocalizations. However, in practice, recording sites also have site-specific variations and unique background audio that need to be removed, warranting the need for site-specific data. Here, we test the potential of training DSSS models on site-specific bird vocalizations and background audio. We used a semiautomated workflow using deep supervised classification and statistical cleaning to label and generate a site-specific source separation dataset by mixing birdsongs and background audio segments. Then, we trained a deep supervised source separation (DSSS) model with this generated dataset. Because most data is passively-recorded and consequently noisy, the true isolated birdsongs are unavailable which makes evaluation challenging. Therefore, in addition to using traditional source separation (SS) metrics, we also show the effectiveness of our site-specific approach using metrics commonly used in ornithological analyses such as automated feature labeling and species-specific trilateration accuracy. Our approach of training on site-specific data boosts the source-to-distortion, source-to-interference, and source-to-artifact ratios (SDR, SIR, and SAR) by 9.33 dB, 24.07 dB, and 3.60 dB respectively. We also find our approach allows for automated feature labeling with single-digit mean absolute percent error and birdsong trilateration accuracy with a mean simulated trilateration error of 2.58m. Overall, we show that site-specific DSSS is a promising upstream solution for wildlife audio analysis tools that break down in the presence of background noise. By training on site-specific data, our method is robust to unique, site-specific interference that caused previous methods to fail.