Domain adaptation algorithms are useful when the distributions of the training and the test data are different. In this paper, we focus on the problem of instrumental variation and time-varying drift in the field of sensors and measurement, which can be viewed as discrete and continuous distributional change in the feature space. We propose maximum independence domain adaptation (MIDA) and semi-supervised MIDA to address this problem. Domain features are first defined to describe the background information of a sample, such as the device label and acquisition time. Then, MIDA learns a subspace which has maximum independence with the domain features, so as to reduce the interdomain discrepancy in distributions. A feature augmentation strategy is also designed to project samples according to their backgrounds so as to improve the adaptation. The proposed algorithms are flexible and fast. Their effectiveness is verified by experiments on synthetic datasets and four real-world ones on sensors, measurement, and computer vision. They can greatly enhance the practicability of sensor systems, as well as extend the application scope of existing domain adaptation algorithms by uniformly handling different kinds of distributional change.

Download full-text PDF

Source
http://dx.doi.org/10.1109/TCYB.2016.2633306DOI Listing

Publication Analysis

Top Keywords

domain features
12
domain adaptation
12
adaptation algorithms
8
sensors measurement
8
distributional change
8
maximum independence
8
independence domain
8
domain
6
learning domain-invariant
4
domain-invariant subspace
4

Similar Publications

Organismal complexity strongly correlates with the number of protein families and domains.

Proc Natl Acad Sci U S A

February 2025

Duncan and Nancy MacMillan Cancer Immunology and Metabolism Center of Excellence, Rutgers Cancer Institute of New Jersey, New Brunswick, NJ 08901.

In the pregenomic era, scientists were puzzled by the observation that haploid genome size (the C-value) did not correlate well with organismal complexity. This phenomenon, called the "C-value paradox," is mostly explained by the fact that protein-coding genes occupy only a small fraction of eukaryotic genomes. When the first genome sequences became available, scientists were even more surprised by the fact that the number of genes (G-value) was also a poor predictor of complexity, which gave rise to the "G-value paradox.

View Article and Find Full Text PDF

Arabidopsis CIRP1 E3 ligase modulates drought and oxidative stress tolerance and reactive oxygen species homeostasis by directly degrading catalases.

J Integr Plant Biol

January 2025

Key Laboratory of National Forestry and Grassland Administration on Plant Conservation and Utilization in Southern China & Guangdong Provincial Key Laboratory of Applied Botany, South China Botanical Garden, Chinese Academy of Sciences, Guangzhou, 510650, China.

Reactive oxygen species (ROS) plays critical roles in modulating plant growth and stress response and its homeostasis is fine tuned using multiple peroxidases. HO, a major kind of ROS, is removed rapidly and directly using three catalases, CAT1, CAT2, and CAT3, in Arabidopsis. Although the activity regulations of catalases have been well studied, their degradation pathway is less clear.

View Article and Find Full Text PDF

Although much has been learned about the entry mechanism of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), many details of the entry mechanisms of seasonal human coronaviruses (HCoVs) remain less well understood. In the present study, we used 293T cell lines stably expressing angiotensin converting enzyme (ACE2), aminopeptidase N (APN), or transmembrane serine protease 2 (TMPRSS2), which support high-level transduction of lentiviral pseudoviruses bearing spike proteins of seasonal HCoVs, HCoV-NL63, -229E, or -HKU1, respectively, to compare spike processing and virus entry pathways among these viruses. Our results showed that the entry of HCoV-NL63, -229E, and -HKU1 pseudoviruses into cells is sensitive to endosomal acidification inhibitors (chloroquine and NHCl), indicating entry via the endocytosis route.

View Article and Find Full Text PDF

The development of chiral organic materials with strong non-reciprocal chiroptical features may have major implications for cutting-edge technological applications. In this work, a new synthesized chiral 1,4-diketo-3,6-dithienylpyrrolo[3,4-]pyrrole dye, bearing two ()-3,7-dimethyl-1-octyl alkyl chains on the lactam moieties and functionalized with two lateral 9-anthracenyl π-conjugated units, exhibited strong non-reciprocal chiroptical properties in thin films, with some important differences between samples prepared by drop casting and spin coating. A detailed study was performed to unravel the intimate structure-property relationship, involving computational analysis, different microscopy techniques and synchrotron radiation Mueller matrix polarimetry imaging (SR-MMP) investigation.

View Article and Find Full Text PDF

The growing demand for biological products drives many efforts to maximize expression of heterologous proteins. Advances in high-throughput sequencing can produce data suitable for building sequence-to-expression models with machine learning. The most accurate models have been trained on one-hot encodings, a mechanism-agnostic representation of nucleotide sequences.

View Article and Find Full Text PDF

Want AI Summaries of new PubMed Abstracts delivered to your In-box?

Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!