Data-dependent Methods Research Articles

Data Independent Acquisition Based Bi-Directional Deep Networks for Biometric ECG Authentication

In this report, the study of non-fiducial based approaches for Electrocardiogram(ECG) biometric authentication is examined, and several excessive techniques are proposed to perform comparative experiments for evaluating the best possible approach for all the classification tasks. Non-fiducial methods are designed to extract the discriminative information of a signal without annotating fiducial points. However, this process requires peak detection to identify a heartbeat signal. Based on recent studies that usually rely on heartbeat segmentation, QRS detection is required, and the process can be complicated for ECG signals for which the QRS complex is absent. Thus, many studies only conduct biometric authentication tasks on ECG signals with QRS complexes, and are hindered by similar limitations. To overcome this issue, we proposed a data-independent acquisition method to facilitate highly generalizable signal processing and feature learning processes. This is achieved by enhancing random segmentation to avoid complicated fiducial feature extraction, along with auto-correlation to eliminate the phase difference due to random segmentation. Subsequently, a bidirectional recurrent neural network (RNN) with long short-term memory (LSTM) deep networks is utilized to automatically learn the features associated with the signal and to perform an authentication task. The experimental results suggest that the proposed data-independent approach using a BLSTM network achieves a relatively high classification accuracy for every dataset relative to the compared techniques. Moreover, it exhibited a significantly higher accuracy rate in experiments using ECG signals without the QRS complex. The results also revealed that data-dependent methods can only perform well for specified data types and amendments of data variations, whereas the presented approach can also be considered for generalization to other quasi-periodical biometric signal-based classification tasks in future studies.

Data-Dependent Feature Extraction Method Based on Non-Negative Matrix Factorization for Weakly Supervised Domestic Sound Event Detection

In this paper, feature extraction methods are developed based on the non-negative matrix factorization (NMF) algorithm to be applied in weakly supervised sound event detection. Recently, the development of various features and systems have been attempted to tackle the problems of acoustic scene classification and sound event detection. However, most of these systems use data-independent spectral features, e.g., Mel-spectrogram, log-Mel-spectrum, and gammatone filterbank. Some data-dependent feature extraction methods, including the NMF-based methods, recently demonstrated the potential to tackle the problems mentioned above for long-term acoustic signals. In this paper, we further develop the recently proposed NMF-based feature extraction method to enable its application in weakly supervised sound event detection. To achieve this goal, we develop a strategy for training the frequency basis matrix using a heterogeneous database consisting of strongly- and weakly-labeled data. Moreover, we develop a non-iterative version of the NMF-based feature extraction method so that the proposed feature extraction method can be applied as a part of the model structure similar to the modern “on-the-fly” transform method for the Mel-spectrogram. To detect the sound events, the temporal basis is calculated using the NMF method and then used as a feature for the mean-teacher-model-based classifier. The results are improved for the event-wise post-processing method. To evaluate the proposed system, simulations of the weakly supervised sound event detection were conducted using the Detection and Classification of Acoustic Scenes and Events 2020 Task 4 database. The results reveal that the proposed system has F1-score performance comparable with the Mel-spectrogram and gammatonegram and exhibits 3–5% better performance than the log-Mel-spectrum and constant-Q transform.

Missing Value Monitoring to Address Missing Values in Quantitative Proteomics.

Many classes of key functional proteins such as transcription factors or cell cycle proteins are present in the proteome at a very low concentration. These low-abundance proteins are almost entirely invisible to systematic quantitative analysis by classical data dependent proteomics methods (DDA). Moreover, DDA runs in shotgun proteomics experiments are plenty of missing values among the replicates due to the stochastic nature of the acquisition method, thus hampering the robustness of the quantitative analysis. Here, we have overcome these obstacles designing a robust workflow named missing value monitoring (MvM) in order to follow low abundance proteins dynamics.

Distributed Supervised Discrete Hashing With Relaxation

The data-dependent hash methods are becoming more and more attractive because they perform well in fast retrieval and storing high-dimensional data. Most existing supervised hashes are centralized, such as supervised discrete hashing (SDH) and supervised discrete hashing with relaxation (SDHR). The SDH algorithm determines the regression target by using ordinary least squares regression and the traditional zero-one matrix encoding of class label information. And SDHR is a constraint to the regression target matrix so that each example is correctly classified and satisfies a larger margin, so as to achieve the purpose of optimizing the regression target. In real environment, a large amount of data will be distributed in different nodes. Therefore, the centralized hash method has great limitations. In this article, we propose distributed supervised discrete hashing algorithm with relaxation (DSDHR) based on SDHR. The SDHR algorithm is introduced into the distributed network. In this framework, all nodes share a centralized hash learning model. At the same time, in order to ensure that the distributed hash algorithm is updated in parallel on multiple nodes, consistency constraints are introduced. In each node, alternative iterative methods are used to obtain the binary hash code, regression target and hash function. Experiments show that DSDHR has certain competitive advantages over some centralized hashing algorithms and some distributed hashing algorithms.

Profiling of Low-Molecular-Weight Carbonyls and Protein Modifications in Flavored Milk.

Thermal treatments of dairy products favor oxidations, Maillard reactions, and the formation of sugar or lipid oxidation products. Additives including flavorings might enhance these reactions or even induce further reactions. Here we aimed to characterize protein modifications in four flavored milk drinks using samples along the production chain—raw milk, pasteurization, mixing with flavorings, heat treatment, and the commercial product. Therefore, milk samples were analyzed using a bottom up proteomics approach and a combination of data-independent (MSE) and data-dependent acquisition methods (DDA). Twenty-one small carbonylated lipids were identified by shotgun lipidomics triggering 13 protein modifications. Additionally, two Amadori products, 12 advanced glycation end products (AGEs), and 12 oxidation-related modifications were targeted at the protein level. The most common modifications were lactosylation, formylation, and carboxymethylation. The numbers and distribution of modification sites present in raw milk remained stable after pasteurization and mixing with flavorings, while the final heat treatment significantly increased lactosylation and hexosylation in qualitative and quantitative terms. The processing steps did not significantly affect the numbers of AGE-modified, oxidized/carbonylated, and lipid-carbonylated sites in proteins.

From 1 to N: A computer-aided case study of thermodynamic cycle construction based on thermodynamic process combination

Thermodynamic cycles are essential for utilization of energy in most situations. The development of methods that can be used to construct thermodynamic cycles has therefore been the focus of engineers and scientists for some time. The essence of most traditional construction methods is to improve the classical thermodynamic cycles rather than constructing a new cycle. Few researchers have attempted to construct cycles with thermodynamic processes as basic elements. In this study, the thermodynamic process is considered as the degree of freedom in thermodynamic cycle construction for the first time, and a computer-aided method for constructing thermodynamic cycles is proposed based on this viewpoint. This method constructs a thermodynamic cycle by combining a specific number of thermodynamic processes under the guidance of predefined rules. Case studies show that this method can generate numerous forms of thermodynamic cycles, including existing classical cycles and new cycles. Thirty-eight new cycles that were generated using this method demonstracted efficiencies higher than 0.25, which is 16.67% lower than the efficiency of the Carnot cycle under the same conditions. This shows that this method is an efficient means of constructing thermodynamic cycles. The method also provides a new perspective on the construction of the thermodynamic cycle, facilitating the transition from empirical methods to data-dependent methods for thermodynamic cycle construction.

Probing SWATH-MS as a tool for proteome level quantification in a nonmodel fish.

Quantitative proteomics via mass spectrometry can provide valuable insight into molecular and phenotypic characteristics of a living system. Recent mass spectrometry developments include data‐independent acquisition (SWATH/DIA‐MS), an accurate, sensitive and reproducible method for analysing the whole proteome. The main requirement for this method is the creation of a comprehensive spectral library. New technologies have emerged producing larger and more accurate species‐specific libraries leading to a progressive collection of proteome references for multiple molecular model species. Here, for the first time, we set out to compare different spectral library constructions using multiple tissues from a coral reef fish to demonstrate its value and feasibility for nonmodel organisms. We created a large spectral library composed of 12,553 protein groups from liver and brain tissues. Via identification of differentially expressed proteins under fish exposure to elevated pCO2 and temperature, we validated the application and usefulness of these different spectral libraries. Successful identification of significant differentially expressed proteins from different environmental exposures occurred using the library with a combination of data‐independent and data‐dependent acquisition methods as well as both tissue types. Further analysis revealed expected patterns of significantly up‐regulated heat shock proteins in a dual condition of ocean warming and acidification indicating the biological accuracy and relevance of the method. This study provides the first reference spectral library for a nonmodel organism. It represents a useful guide for future building of accurate spectral library references in nonmodel organisms allowing the discovery of ecologically relevant changes in the proteome.

Unsupervised Ensemble Hashing: Boosting Minimum Hamming Distance

Hashing aims at learning discriminative binary codes of high-dimensional data for the approximate nearest neighbor searching. However, the distance ranking obtained by traditional methods is not optimum in the Hamming space, and it degrades the performance for retrieval tasks. To tackle the above problem, an unsupervised ensemble hashing is proposed to improve the ranking accuracy by assembling the diverse hash tables independently in this paper. We observe that the higher the accuracy is the larger diversity the base learner has, and the more effective the ensemble method is. Based on this principle, two special ensembles hashing approaches are proposed to increase diversity by bootstrap sampling with data-dependent methods. Especially, the results are better when the minimum Hamming distance is large and the variance of the Hamming distance is small. This proposed method is conducted in the experiments and the results show that it can achieve about 10%-25% performance compared with the baseline algorithm, which achieves competitive results with the state-of-the-art methods on the CIFAR-10 and LabelMe benchmarks.

Improving Proteomics Data Reproducibility with a Dual-Search Strategy.

Mass spectrometry-based proteomics is an invaluable tool for addressing important biological questions. Data-dependent acquisition methods effectuate stochastic acquisition of data in complex mixtures, which results in missing identifications across replicates. We developed a search approach that improves the reproducibility of data acquired from any mass spectrometer. In our approach, a spectral library is built from the identification results from a database search, and then, the library is used to research the same data files to obtain the final result. We showed that higher identification and quantification reproducibility is achieved with the dual-search approach than with a typical database search. Four datasets with different complexity were compared: (1) data from a cell lysate study performed in our lab, (2) data from an interactome study performed in our lab, (3) a publicly available extracellular vesicles dataset, and (4) a publicly available phosphoproteomics dataset. Our results show that the dual-search approach can be widely and easily used to improve data quality in proteomics data.

Descriptors for Electrolyte-Renormalized Oxidative Stability of Solvents in Lithium-Ion Batteries.

Electrolyte stability against oxidation is one of the important factors limiting the development of high energy density batteries. The highest occupied molecular orbital (HOMO) level of solvent molecules has been traditionally used for understanding trends in their oxidative stability, but this assumes a noninteracting environment. However, solvent HOMO levels are renormalized because of molecules in their solvation shells. In this work, we first demonstrate an inexpensive and accurate method to determine the HOMO level of the solvent followed by simple descriptors for renormalization of the HOMO level due to different electrolyte components. The descriptors are based on Gutmann donor and acceptor numbers of the solvent and other components. The method uses fast generalized gradient approximation-level density functional theory calculations compared to previously used expensive, experimental data-dependent methods. This method can be used to screen for unexplored stable solvents among the large number of known organic compounds to design novel high-voltage stable electrolytes.

The sensitivity of unit root tests to the initial condition and to the lag length selection: A Monte Carlo Simulation Study

While the recent literature has discussed the effect of the deviation of the initial observation of the economic series from its deterministic component (initial condition) on unit root tests, no studies have examined to date the effect of the selection of the lag length on unit root tests in this setting. Our study aims to fill this gap, and provides a recommendation for the practitioner. The objective is to investigate to what extent the sensitivity of the outcome of unit root tests to the initial condition changes with the use of both standard and modified data-dependent methods to select the lag length in the augmented autoregression, even for those tests that have been considered robust in the presence of uncertainty about the initial condition. To do so, we conduct a Monte Carlo simulation study to analyse the finite sample properties (size and power) of unit root tests based on alternative lag selection criteria and different magnitudes of the initial condition.

A Data-Independent Reusable Projection (DIRP) Technique for Dimension Reduction in Big Data Classification Using k-Nearest Neighbor (k-NN)

As the high-dimensional data impact the performance of the k-Nearest Neighbor (k-NN) classification algorithm, dimension reduction for the big data classification using k-NN algorithm draws huge attention from the industry as well as academia. The popular dimension reduction techniques such as PCA, LDA, SVD are data-dependent methods where the projection matrix was generated from the input data which make the reusability of the projection matrix impractical. In this paper, a Data-Independent Reusable Projection (DIRP) technique has been proposed to project high-dimensional data to low-dimensional data and prove how the projection matrix can be reused for any dataset with same number of dimensions. The proposed DIRP method preserves the distance between any two points in the dataset which works well for the distance-based classification algorithms like k-NN. The DIRP method has been implemented in R, and a new package “RandPro” for generating projection matrix has been developed and tested with the CIFAR-10, handwritten digit recognition (MNIST) dataset and human activity recognition dataset. The two versions of the RandPro package have been uploaded in the CRAN repository. The running time and classification metrics comparison between PCA and DIRP method has been analyzed. From the results, it has been found that the running time of the proposed method is reduced significantly with the near equivalent accuracy to the original data.

The importance of interpretability and visualization in machine learning for applications in medicine and health care

In a short period of time, many areas of science have made a sharp transition towards data-dependent methods. In some cases, this process has been enabled by simultaneous advances in data acquisition and the development of networked system technologies. This new situation is particularly clear in the life sciences, where data overabundance has sparked a flurry of new methodologies for data management and analysis. This can be seen as a perfect scenario for the use of machine learning and computational intelligence techniques to address problems in which more traditional data analysis approaches might struggle. But, this scenario also poses some serious challenges. One of them is model interpretability and explainability, especially for complex nonlinear models. In some areas such as medicine and health care, not addressing such challenge might seriously limit the chances of adoption, in real practice, of computer-based systems that rely on machine learning and computational intelligence methods for data analysis. In this paper, we reflect on recent investigations about the interpretability and explainability of machine learning methods and discuss their impact on medicine and health care. We pay specific attention to one of the ways in which interpretability and explainability in this context can be addressed, which is through data and model visualization. We argue that, beyond improving model interpretability as a goal in itself, we need to integrate the medical experts in the design of data analysis interpretation strategies. Otherwise, machine learning is unlikely to become a part of routine clinical and health care practice.

Partial randomness hashing applied to remote sensing object classification

Recently, object classification of remote sensing images has attracted more and more research interests due to the development of satellite and aerial vehicle technologies. Hashing learning is an efficient method to handle the huge amount of the remote sensing data. We proposed a hashing learning method named partial randomness supervised discrete hashing (PRSDH), which combines data-dependent methods and data-independent methods. It jointly learns a discrete binary codes generation and partial random constraint optimization model. By random projection, the computation complexity is reduced effectively. With the weight matrix derived from the training data, the semantic similarity between the data can be well preserved while generating the hashing codes. For the discrete constraint problem, this paper adopts the discrete cyclic coordinate descent algorithm to optimize the codes bit by bit. The experimental results show that PRSDH outperforms other comparative methods and demonstrate that PRSDH has good adaptability to the characteristic of remote sensing object.

Risk consistency of cross-validation with Lasso-type procedures

The lasso and related sparsity inducing algorithms have been the target of substantial theoretical and applied research. Correspondingly, many results are known about their behavior for a fixed or optimally chosen tuning parameter specified up to unknown constants. In practice, however, this oracle tuning parameter is inaccessible so one must use the data to select one. Common statistical practice is to use a variant of cross-validation for this task. However, little is known about the theoretical properties of the resulting predictions with such data-dependent methods. We consider the high-dimensional setting with random design wherein the number of predictors $p$ grows with the number of observations $n$. Under typical assumptions on the data generating process, similar to those in the literature, we recover oracle rates up to a log factor when choosing the tuning parameter with cross-validation. Under weaker conditions, when the true model is not necessarily linear, we show that the lasso remains risk consistent relative to its linear oracle. We also generalize these results to the group lasso and square-root lasso and investigate the predictive and model selection performance of cross-validation via simulation.

Targeted MultiNotch MS3 Approach for Relative Quantification of N-Glycans Using Multiplexed Carbonyl-Reactive Isobaric Tags.

The recently developed and commercially available carbonyl-reactive tandem mass tags (aminoxyTMT) enable multiplexed quantification of glycans through comparison of reporter ion intensities. However, challenges still exist for collision activated dissociation (CAD) MS/MS based quantification of aminoxyTMT due to the relatively low reporter ion yield especially for glycans with labile structures. To circumvent this limitation, we utilized the unique structural features of N-glycan molecules, the common core sugar sequence (HexNAc)2(Man)3, and common m/z of Yn ions generated from different types of precursors by MS/MS and designed a Y1 ion triggered, targeted MultiNotch MS3 relative quantification approach based on aminoxyTMT labeling. This approach was implemented on a nanoHILIC-Tribrid quadrupole-ion trap-Orbitrap platform, which enables prescreening of aminoxyTMT labeled N-glycan precursor ions by Y1 ion fragment ion mass in a higher-energy collisional dissociation (HCD) MS/MS scan and coisolation and cofragmentation of multiple Yn fragment ions that carry the isobaric tags from the inclusion list in the MS/MS/MS scan. Through systematical optimization and evaluation using N-glycans released from several glycoprotein standards and human serum proteins, we demonstrated that the Y1 ion triggered, targeted MultiNotch MS3 approach offers improved accuracy, precision, and sensitivity for relative quantification compared to traditional data-dependent MS2 and Y1 ion MS3 quantification methods.

Empirical null estimation using zero-inflated discrete mixture distributions and its application to protein domain data.

In recent mutation studies, analyses based on protein domain positions are gaining popularity over gene-centric approaches since the latter have limitations in considering the functional context that the position of the mutation provides. This presents a large-scale simultaneous inference problem, with hundreds of hypothesis tests to consider at the same time. This article aims to select significant mutation counts while controlling a given level of Type I error via False Discovery Rate (FDR) procedures. One main assumption is that the mutation counts follow a zero-inflated model in order to account for the true zeros in the count model and the excess zeros. The class of models considered is the Zero-inflated Generalized Poisson (ZIGP) distribution. Furthermore, we assumed that there exists a cut-off value such that smaller counts than this value are generated from the null distribution. We present several data-dependent methods to determine the cut-off value. We also consider a two-stage procedure based on screening process so that the number of mutations exceeding a certain value should be considered as significant mutations. Simulated and protein domain data sets are used to illustrate this procedure in estimation of the empirical null using a mixture of discrete distributions. Overall, while maintaining control of the FDR, the proposed two-stage testing procedure has superior empirical power.

A global Staphylococcus aureus proteome resource applied to the in vivo characterization of host-pathogen interactions

Data-independent acquisition mass spectrometry promises higher performance in terms of quantification and reproducibility compared to data-dependent acquisition mass spectrometry methods. To enable high-accuracy quantification of Staphylococcus aureus proteins, we have developed a global ion library for data-independent acquisition approaches employing high-resolution time of flight or Orbitrap instruments for this human pathogen. We applied this ion library resource to investigate the time-resolved adaptation of S. aureus to the intracellular niche in human bronchial epithelial cells and in a murine pneumonia model. In epithelial cells, abundance changes for more than 400 S. aureus proteins were quantified, revealing, e.g., the precise temporal regulation of the SigB-dependent stress response and differential regulation of translation, fermentation, and amino acid biosynthesis. Using an in vivo murine pneumonia model, our data-independent acquisition quantification analysis revealed for the first time the in vivo proteome adaptation of S. aureus. From approximately 2.15 × 105 S. aureus cells, 578 proteins were identified. Increased abundance of proteins required for oxidative stress response, amino acid biosynthesis, and fermentation together with decreased abundance of ribosomal proteins and nucleotide reductase NrdEF was observed in post-infection samples compared to the pre-infection state.

Regularizing Neural Networks via Retaining Confident Connections

Regularization of neural networks can alleviate overfitting in the training phase. Current regularization methods, such as Dropout and DropConnect, randomly drop neural nodes or connections based on a uniform prior. Such a data-independent strategy does not take into consideration of the quality of individual unit or connection. In this paper, we aim to develop a data-dependent approach to regularizing neural network in the framework of Information Geometry. A measurement for the quality of connections is proposed, namely confidence. Specifically, the confidence of a connection is derived from its contribution to the Fisher information distance. The network is adjusted by retaining the confident connections and discarding the less confident ones. The adjusted network, named as ConfNet, would carry the majority of variations in the sample data. The relationships among confidence estimation, Maximum Likelihood Estimation and classical model selection criteria (like Akaike information criterion) is investigated and discussed theoretically. Furthermore, a Stochastic ConfNet is designed by adding a self-adaptive probabilistic sampling strategy. The proposed data-dependent regularization methods achieve promising experimental results on three data collections including MNIST, CIFAR-10 and CIFAR-100.

SWATH-ID: An instrument method which combines identification and quantification in a single analysis.

Data-independent acquisition (DIA) approaches, such as SWATH® -MS, are showing great potential to reliably quantify significant numbers of peptides and proteins in an unbiased manner. These developments have enhanced interest in developing a single DIA method that integrates qualitative and quantitative analysis, eliminating the need of a prebuilt library of peptide spectra, which are created through data-dependent acquisition methods or from public repositories. Here, we introduce a new DIA approach, referred to as "SWATH-ID," which was developed to allow peptide identification as well as quantitation. The SWATH-ID method is composed of small Q1 windows, achieving better selectivity and thus significantly improving high-confidence peptide extractions from data files. Furthermore, the SWATH-ID approach transmits precursor ions without fragmentation as well as their fragments within the same SWATH acquisition period. This provides a single scan that includes all precursor ions within the isolation window as well as a record of all of their fragment ions, substantially negating the need for a survey scan. In this way all precursors present in a small Q1 window are associated with their fragment ions, improving the identification specificity and providing a more comprehensive and in-depth view of protein and peptide species in complex samples.