Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “label shift”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 19 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

High dimensional binary classification under label shift: phase transition and regularization

Label Shift has been widely believed to be harmful to the generalization performance of machine learning models. Researchers have proposed many approaches to mitigate the impact of the label shift, e.g., balancing the training data. However, these methods often consider the underparametrized regime, where the sample size is much larger than the data dimension. The research under the overparametrized regime is very limited. Here, to bridge this gap, we propose a new asymptotic analysis of the Fisher Linear Discriminant classifier for binary classification with label shift. Specifically, we prove that there exists a phase transition phenomenon: Under certain overparametrized regime, the classifier trained using imbalanced data outperforms the counterpart with reduced balanced data. Moreover, we investigate the impact of regularization to the label shift: The aforementioned phase transition vanishes as the regularization becomes strong.

binary classification↗
DOE OSTI2024

Assessing Membership Inference Attacks under Distribution Shifts

Membership inference attacks (MIAs) exploit machine learning models to infer whether a data point was in the training set, posing significant privacy risks even with limited black-box access. These attacks rely on the attacker approximating the target model’s training distribution, yet the impact of distribution shifts between target and shadow models on MIA success remains underexplored. We systematically evaluate five types of distribution shifts —-cutout, jitter, Gaussian noise, label shift, and attribute shift —- at varying intensities. Our results reveal that these shifts affect MIA effectiveness in nuanced ways, with some reducing attack success while others exacerbate vulnerabilities, and the same shift can have opposite effects depending on the type of MIA. This highlights the complex interplay between distributional differences and attack performance, offering critical insights for improving model defenses against MIAs.

Shi, Yichuan [Massachusetts Institute of Technolog↗
DOE OSTI2024

Boundary-Aware Adversarial Learning Domain Adaption and Active Learning for Cross-Sensor Building Extraction

The use of convolutional neural networks (CNNs) for building extraction from remote sensing images has been widely studied and many public datasets have been made available for accelerating development of these CNN models. Yet adapting pretrained models at scale in real-world scenarios remains a challenging task. The main barrier is that certain new labels are still needed to compensate for domain shifting between the labeled data and new images that potentially cover new geographic locations or that are from a different sensor. In this article, we propose to add informatively labeled samples from a new image pool under the paradigm of active learning. To select the most useful samples based on model uncertainty, we first tackle the problem of uncalibrated uncertainty estimation due to distribution shifting by adapting feature extractors with boundary-based adversarial learning. Calibrated uncertainty is used as the query criterion in the active learning process, where the most uncertain samples are selected for annotation and included for model retraining. The proposed workflow was tested with three data pairs in which each workflow represents a scenario often encountered in real-world applications, including adapting pretrained models to new images collected with different sensors or to new geographic areas where appearances and types of buildings are very different. Compared to several baselines, including random sampling, temperature scaling (a well-known uncertainty calibration technique), different query strategies, and active domain adaptation methods, the proposed workflow shows that strategically querying a smaller set of samples for labeling achieves comparable or better building extraction performance. The proposed method reduces the number of labeled samples required to achieve sufficient model accuracy, thus significantly reducing hundreds of person-hours for labeled data creation. In addition, we include a few considerations when deploying this workflow in a GPU cluster that can be easily adapted to achieve operational building extraction model retraining.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2013

Redox states of Desulfovibrio vulgaris DsrC, a key protein in dissimilatory sulfite reduction

Highlights: •DsrC is known to interact with the dissimilatory sulfite reductase enzyme (DsrAB). •We show that, however, most cellular DsrC is not associated with DsrAB. •A gel-shift assay was developed that allows monitoring of the DsrC redox state. •The DsrC intramolecularly oxidized state could only be produced by arginine treatment. -- Abstract: Dissimilatory reduction of sulfite is carried out by the siroheme enzyme DsrAB, with the involvement of the protein DsrC, which has two conserved redox-active cysteines. DsrC was initially believed to be a third subunit of DsrAB. Here, we report a study of the distribution of DsrC in cell extracts to show that, in the model sulfate reducer Desulfovibrio vulgaris, the majority of DsrC is not associated with DsrAB and is thus free to interact with other proteins. In addition, we developed a cysteine-labelling gel-shift assay to monitor the DsrC redox state and behaviour, and procedures to produce the different redox forms. The oxidized state of DsrC with an intramolecular disulfide bond, which is proposed to be a key metabolic intermediate, could be successfully produced for the first time by treatment with arginine.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2010

Isotopic Effect on Ion Mobility and Separation of Isotopomers by High-Field Ion Mobility Spectrometry

Since early 1900-s, when vacuum techniques and ion detectors first enabled investigations of gas-phase ions, two approaches to their separation and characterization have emerged - mass spectrometry (MS) and ion mobility spectrometry (IMS).1,2 Though both exploit that distinct charged species move in electric fields differently, MS is performed in vacuum and is based only on the ion mass/charge (m/q) ratio while IMS involves sufficiently dense buffer gases and relies on ion transport properties. The first major discovery enabled by MS was the existence of isotopes by Thomson and Aston,3 and isotopic analyses have since been integral to MS. In particular, the preparative separation of U isotopes using Lawrence’s Calutron was the first industrial application of MS,4 and isotopic labeling is key to MS quantification methods. With IMS, the issue of isotopes was largely ignored as the resolving power (R) was generally too low for their separation. Here, we demonstrate that recently developed high-resolution differential IMS can separate isotopic molecular ions, including nominal isobars with different isotopic content and isotopomers. This capability may enable a new method for isotope separation in a small-scale format at ambient pressure and aid localization of labeled sites in various molecules. Perhaps most importantly, the isotopic shifts depend on the labeled atom position and thus may contain the kind of detailed structural information that is available in solution or solid state using tools such as NMR but has not generally been obtainable for gas-phase ions.

38 RADIATION CHEMISTRY, RADIOCHEMISTRY, AND NUCLEA↗
DOE OSTI2011

Trypsin-catalyzed oxygen-18 labeling for quantitative proteomics

Stable isotope labeling based on relative peptide/protein abundance measurements is commonly applied for quantitative proteomics. Recently, trypsin-catalyzed oxygen-18 labeling has grown in popularity due to its simplicity, cost-effectiveness, and its ability to universally label peptides with high sample recovery. In (18)O labeling, both C-terminal carboxyl group atoms of tryptic peptides can be enzymatically exchanged with (18)O, thus providing the labeled peptide with a 4 Da mass shift from the (16)O-labeled sample. Peptide (18)O labeling is ideally suited for generating a labeled "universal" reference sample used for obtaining accurate and reproducible quantitative measurements across large number of samples in quantitative discovery proteomics.

18O labeling↗
DOE OSTI2022

Automated Shift Detection in Sensor-Based PV Power and Irradiance Time Series: Preprint

PV power and irradiance sensor-based measurements are prone to error, resulting in issues such as abrupt time series data shifts. These shifts, which are usually unintentional, may be caused by software or hardware configuration changes on a PV system, and do not reflect an actual change in overall system performance. Locating these shifts and segmenting the associated time series aids in more accurate future PV analysis. In this research, an offline changepoint detection (CPD) algorithm that automatically detects these abrupt data shifts in sensor-based time series is introduced. Data shift periods in 101 daily PV power and irradiance time series were labeled manually by two solar experts. These data streams represent sensor-based measurements, and display a variety of data shift behaviors. A changepoint detection algorithm was tuned using the 101 labeled data streams, with each model configuration's ability to detect labeled changepoints benchmarked using metrics such as F1-score, recall, and Rand Index. Best performing models on seasonality-corrected data streams include the Pruned Exact Linear (PELT) method, the Binary Segmentation method, and the Bottom-Up method, all scoring an average F1-score of 0.76 or greater at detecting labeled changepoints within a 30-day window for the labeled data sets. To promote further research in this space, we are releasing the labeled data shift sets on U.S. Department of Energy's (DOE) DuraMAT Data Hub, and the associated algorithm in the Python PVAnalytics package.

changepoint detection↗
DOE OSTI2022

Automated Shift Detection in Sensor-Based PV Power and Irradiance Time Series

PV power and irradiance sensor-based measurements are prone to error, resulting in issues such as abrupt time series data shifts. These shifts, which are usually unintentional, may be caused by software or hardware configuration changes on a PV system, and do not reflect an actual change in overall system performance. Locating these shifts and segmenting the associated time series aids in more accurate future PV analysis. In this research, an offline changepoint detection (CPD) algorithm that automatically detects these abrupt data shifts in sensor-based time series is introduced. Data shift periods in 101 daily PV power and irradiance time series were labeled manually by two solar experts. These data streams represent sensor-based measurements, and display a variety of data shift behaviors. A changepoint detection algorithm was tuned using the 101 labeled data streams, with each model configuration's ability to detect labeled changepoints benchmarked using metrics such as F1-score, recall, and Rand Index. Best performing models on seasonality-corrected data streams include the Pruned Exact Linear (PELT) method, the Binary Segmentation method, and the Bottom-Up method, all scoring an average F1-score of 0.76 or greater at detecting labeled changepoints within a 30-day window for the labeled data sets. To promote further research in this space, we are releasing the labeled data shift sets on U.S. Department of Energy's (DOE) DuraMAT Data Hub, and the associated algorithm in the Python PVAnalytics package.

changepoint detection↗
DOE OSTI2013

Stable isotope-assisted NMR characterization of interaction between lipid A and sarcotoxin IA, a cecropin-type antibacterial peptide

Highlights: ► Recombinant sarcotoxin IA was successfully produced with {sup 13}C- and {sup 15}N-labeling. ► Sarcotoxin IA adopts an N-terminal α-helix upon binding to lipid A-embedding micelles. ► Two lysine residues are involved in lipid A-mediated antibacterial activities. -- Abstract: Sarcotoxin IA is a 39-residue cecropin-type peptide from Sarcophaga peregrina. This peptide exhibits antibacterial activity against Gram-negative bacteria through its interaction with lipid A, a core component of lipopolysaccharides. To acquire detailed structural information on this specific interaction, we performed NMR analysis using bacterially expressed sarcotoxin IA analogs with {sup 13}C- and {sup 15}N-labeling along with lipid A-embedding micelles composed of dodecylphosphocholine. By inspecting the stable isotope-assisted NMR data, we revealed that the N-terminal segment (Leu3–Arg18) of sarcotoxin IA formed an amphiphilic α-helix upon its interaction with the aqueous micelles. Furthermore, chemical shift perturbation data indicated that the amino acid residues displayed on this α-helix were involved in the specific interaction with lipid A. On the basis of these data, we successfully identified Lys4 and Lys5 as key residues in the interaction with lipid A and the consequent antibacterial activity. Therefore, these results provide unique information for designing chemotherapeutics based on antibacterial peptide structures.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2019

Spectroscopic Characterization of $μ$-$η$ 1 :$η$ 1 -Peroxo Ligands Formed by Reaction of Dioxygen with Electron-Rich Iridium Clusters

Although oxygen is a common ligand in supported metal catalysts, its coordination has been challenging to elucidate. Herein we characterize a diiridium complex that has been previously shown by X-ray diffraction crystallography to incorporate a μ-η 1 :η 1 -peroxo ligand. We observe markedly enhanced intensity at 788 cm -1 in the Raman spectrum of this complex, which is a consequence of bonding of the peroxo ligand but does not shift upon 18 O labeling. Electronic structure calculations at the density functional theory level suggest that this increase in Raman intensity results from bands associated with rocking of CH 2 substituents directly attached to P(Ph) 2 groups coupling with the O-O band. These results provide part of the foundation for understanding oxygen ligands on a silica-supported tetrairidium carbonyl cluster stabilized with bulky electron-donating phosphine ligands [ p-tert -butyl-calix[4]arene(OPr) 3 (OCH 2 PPh 2 ) (Ph = phenyl; Pr = propyl)]. Reaction of the cluster with O2 also led to the growing in of a Raman band at 788 cm -1 , similar to that in the diiridium complex and also assigned to the bonding of a bridging peroxo ligand. Infrared spectra recorded as the supported cluster reacted in sequential exposures to (i) H 2 , (ii) O 2 , (iii) H 2 , and (iv) CO indicate that two bridging peroxo ligands were bonded irreversibly per tetrairidium cluster, replacing bridging carbonyl ligands without altering either the cluster frame or the phosphine ligands. X-ray absorption near edge and infrared spectra include isosbestic points signifying a stoichiometrically simple reaction of the cluster with O 2 , and mass spectra of the effluent gas show that CO 2 formed by oxidation of one terminal CO ligand per cluster as H 2 (and not H 2 O) formed, evidence that hydride ligands had been present on the cluster following treatment (i). The understanding of how O 2 reacts with the metal polyhedron provides a foundation for understanding of how oxidation catalysis may proceed on the surfaces of noble metals.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Domain Shift Analysis in Chest Radiographs Classification in a Veterans Healthcare Administration Population

This study aims to assess the impact of domain shift on chest X-ray classification accuracy and to analyze the influence of ground truth label quality and demographic factors such as age group, sex, and study year. We used a DenseNet121 model pre-trained MIMIC-CXR dataset for deep learning-based multi-label classification using ground truth labels from radiology reports extracted using the CheXpert and CheXbert Labeler. We compared the performance of the 14 chest X-ray labels on the MIMIC-CXR and Veterans Healthcare Administration chest X-ray dataset (VA-CXR). The validation of ground truth and the assessment of multi-label classification performance across various NLP extraction tools revealed that the VA-CXR dataset exhibited lower disagreement rates than the MIMIC-CXR datasets. Additionally, there were notable differences in AUC scores between models utilizing CheXpert and CheXbert. When evaluating multi-label classification performance across different datasets, minimal domain shift was observed in the unseen VA dataset, except for the label “Enlarged Cardiomediastinum.” The subgroup with the most significant variations in multi-label classification performance was study year. These findings underscore the importance of considering domain shift in chest X-ray classification tasks, paying particular attention to the temporality of the exam. Our study reveals the significant impact of domain shift and demographic factors on chest X-ray classification, emphasizing the need for improved transfer learning and robust model development. Addressing these challenges is crucial for advancing medical imaging research and improving patient care.

chest X-ray image classification↗
DOE OSTI2014

NMDA-induced accumulation of Shank at the postsynaptic density is mediated by CaMKII

Highlights: • NMDA-induces accumulation of Shank at the postsynaptic density. • Shank accumulation is preferential to the distal region of the postsynaptic density. • Shank accumulation is mediated by CaMKII. - Abstract: Shank is a specialized scaffold protein present in high abundance at the postsynaptic density (PSD). Using pre-embedding immunogold electron microscopy on cultured hippocampal neurons, we had previously demonstrated further accumulation of Shank at the PSD under excitatory conditions. Here, using the same experimental protocol, we demonstrate that a cell permeable CaMKII inhibitor, tatCN21, blocks NMDA-induced accumulation of Shank at the PSD. Furthermore we show that NMDA application changes the distribution pattern of Shank at the PSD, promoting a 7–10 nm shift in the median distance of Shank labels away from the postsynaptic membrane. Inhibition of CaMKII with tatCN21 also blocks this shift in the distribution of Shank. Altogether these results imply that upon activation of NMDA receptors, CaMKII mediates accumulation of Shank, preferentially at the distal regions of the PSD complex extending toward the cytoplasm.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2017

Applications of NMR and computational methodologies to study protein dynamics

The overwhelming evidence now illustrates the defining role of atomic-scale protein flexibility in biological events such as allostery, cell signaling, and enzyme catalysis. Over the years, spin relaxation nuclear magnetic resonance (NMR) has provided significant insights on the structural motions occurring on multiple time frames over the course of a protein life span. The present review article aims to illustrate to the broader community how this technique continues to shape many areas of protein science and engineering, in addition to being an indispensable tool for studying atomic-scale motions and functional characterization. Continuing developments in underlying NMR technology alongside software and hardware developments for complementary computational approaches now enable methodologies to routinely provide spatial directionality and structural representations traditionally harder to achieve solely using NMR spectroscopy. In addition to its well-established role in structural elucidation, we present recent examples that illustrate the combined power of selective isotope labeling, relaxation dispersion experiments, chemical shift analyses, and computational approaches for the characterization of conformational sub-states in proteins and enzymes.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2010

Site-specific protein backbone and side-chain NMR chemical shift and relaxation analysis of human vinexin SH3 domain using a genetically encoded {sup 15}N/{sup 19}F-labeled unnatural amino acid

Research highlights: {yields} Chemical synthesis of {sup 15}N/{sup 19}F-trifluomethyl phenylalanine. {yields} Site-specific incorporation of {sup 15}N/{sup 19}F-trifluomethyl phenylalanine to SH3. {yields} Site-specific backbone and side chain chemical shift and relaxation analysis. {yields} Different internal motions at different sites of SH3 domain upon ligand binding. -- Abstract: SH3 is a ubiquitous domain mediating protein-protein interactions. Recent solution NMR structural studies have shown that a proline-rich peptide is capable of binding to the human vinexin SH3 domain. Here, an orthogonal amber tRNA/tRNA synthetase pair for {sup 15}N/{sup 19}F-trifluoromethyl-phenylalanine ({sup 15}N/{sup 19}F-tfmF) has been applied to achieve site-specific labeling of SH3 at three different sites. One-dimensional solution NMR spectra of backbone amide ({sup 15}N){sup 1}H and side-chain {sup 19}F were obtained for SH3 with three different site-specific labels. Site-specific backbone amide ({sup 15}N){sup 1}H and side-chain {sup 19}F chemical shift and relaxation analysis of SH3 in the absence or presence of a peptide ligand demonstrated different internal motions upon ligand binding at the three different sites. This site-specific NMR analysis might be very useful for studying large-sized proteins or protein complexes.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2015

Analysis of Cytochrome P450 CYP119 Ligand-dependent Conformational Dynamics by Two-dimensional NMR and X-ray Crystallography

Defining the conformational states of cytochrome P450 active sites is critical for the design of agents that minimize drug-drug interactions, the development of isoform-specific P450 inhibitors, and the engineering of novel oxidative catalysts. In this paper, we used two-dimensional 1H,15N HSQC chemical shift perturbation mapping of 15N-labeled Phe residues and x-ray crystallography to examine the ligand-dependent conformational dynamics of CYP119. Active site Phe residues were most affected by the binding of azole inhibitors and fatty acid substrates, in agreement with active site localization of the conformational changes. This was supported by crystallography, which revealed movement of the F-G loop with various azoles. Nevertheless, the NMR chemical shift perturbations caused by azoles and substrates were distinguishable. The absence of significant chemical shift perturbations with several azoles revealed binding of ligands to an open conformation similar to that of the ligand-free state. In contrast, 4-phenylimidazole caused pronounced NMR changes involving Phe-87, Phe-144, and Phe-153 that support the closed conformation found in the crystal structure. The same closed conformation is observed by NMR and crystallography with a para-fluoro substituent on the 4-phenylimidazole, but a para-chloro or bromo substituent engendered a second closed conformation. An open conformation is thus favored in solution with many azole ligands, but para-substituted phenylimidazoles give rise to two closed conformations that depend on the size of the para-substituent. Finally, the results suggest that ligands selectively stabilize discrete cytochrome P450 conformational states.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2019

Identification of site-specific isotopic labels by vibrational spectroscopy in the electron microscope

The identification of isotopic labels by conventional macroscopic techniques lacks spatial resolution and requires relatively large quantities of material for measurements. We recorded the vibrational spectra of an α amino acid, l -alanine, with damage-free “aloof” electron energy-loss spectroscopy in a scanning transmission electron microscope to directly resolve carbon-site–specific isotopic labels in real space with nanoscale spatial resolution. An isotopic red shift of 4.8 ± 0.4 milli–electron volts in C–O asymmetric stretching modes was observed for 13 C-labeled l -alanine at the carboxylate carbon site, which was confirmed by macroscopic infrared spectroscopy and theoretical calculations. The accurate measurement of this shift opens the door to nondestructive, site-specific, spatially resolved identification of isotopically labeled molecules with the electron microscope.

74 ATOMIC AND MOLECULAR PHYSICS↗