Learning With Noisy Labels Over Imbalanced Subpopulations.

Learning with noisy labels (LNL) has attracted significant attention from the research community. Many recent LNL methods rely on the assumption that clean samples tend to have a "small loss." However, this assumption often fails to generalize to some real-world cases with imbalanced subpopulations, that is, training subpopulations that vary in sample size or recognition difficulty. Therefore, recent LNL methods face the risk of misclassifying those "informative" samples (e.g., hard samples or samples in the tail subpopulations) into noisy samples, leading to poor generalization performance. To address this issue, we propose a novel LNL method to deal with noisy labels and imbalanced subpopulations simultaneously. It first leverages sample correlation to estimate samples' clean probabilities for label correction and then utilizes corrected labels for distributionally robust optimization (DRO) to further improve the robustness. Specifically, in contrast to previous works using classification loss as the selection criterion, we introduce a feature-based metric that takes the sample correlation into account for estimating samples' clean probabilities. Then, we refurbish the noisy labels using the estimated clean probabilities and the pseudo-labels from the model's predictions. With refurbished labels, we use DRO to train the model to be robust to subpopulation imbalance. Extensive experiments on a wide range of benchmarks demonstrate that our technique can consistently improve state-of-the-art (SOTA) robust learning paradigms against noisy labels, especially when encountering imbalanced subpopulations. We provide our code in https://github.com/chenmc1996/LNL-IS.

SSCpred: Single-Sequence-Based Protein Contact Prediction Using Deep Fully Convolutional Network.

There has been a significant improvement in protein residue contact prediction in recent years. Nevertheless, state-of-the-art methods still show deficiencies in the contact prediction of proteins with low-homology information. These top methods depend largely on statistical features that derived from homologous sequences, but previous studies, along with our analyses, show that they are insufficient for inferencing an accurate contact map for nonhomology protein targets. To compensate, we proposed a brand new single-sequence-based contact predictor (SSCpred) that performs prediction through the deep fully convolutional network (Deep FCN) with only the target sequence itself, i.e., without additional homology information. The proposed pipeline makes good use of the target sequence by utilizing the pair-wise encoding technique and Deep FCN. Experimental results demonstrated that SSCpred can produce accurate predictions based on the efficient pipeline. Compared with several most recent methods, SSCpred achieves completive performance on nonhomology targets. Overall, we explored the possibilities of single-sequence-based contact prediction and designed a novel pipeline without using a complex and redundant feature set. The proposed SSCpred can compensate for current methods' disadvantages and achieves better performance on the nonhomology targets. The web server of SSCpred is freely available at http://csbio.njust.edu.cn/bioinf/sscpred/.

A soluble star-shaped silsesquioxane-cored polymer-towards novel stabilization of pH-dependent high internal phase emulsions.

A well-defined pH-responsive star-shaped polymer containing poly(N,N-dimethylaminoethyl methacrylate) (PDMA) arms and a cage-like methacryloxypropyl silsesquioxane (CMSQ-T10) core was used as an interfacial stabilizer for emulsions consisting of m-xylene and water. We explored the properties of the CMSQ/PDMA star-shaped polymer using the characteristic results of nuclear magnetic resonance (NMR) spectroscopy, size exclusion chromatography (SEC), dynamic light scattering (DLS), and zeta potential and conductivity measurements. The interfacial tension results showed that the CMSQ/PDMA star-shaped polymer reduced the interfacial tension between water and oil in a pH-dependent manner. Gelled high internal phase emulsions (HIPEs) including o/w and w/o types were formed in the pH ranges of 1.2-5.8 and 9.1-12.3 with the CMSQ/PDMA star-shaped polymer as a stabilizer, when the oil fractions were 80-90 vol% and 10-20 vol%, respectively. The soluble star-shaped polymer aggregated spontaneously to form a microgel that adsorbed to the two immiscible phases. Images of the fluorescently labeled polymers demonstrated that there was a star-shaped polymer in the continuous phase, and the non-Pickering stabilization based on the percolating network of the star-shaped polymer also contributed to the stabilization of the HIPE. This pH-dependent HIPE was prepared with a novel stabilization mechanism consisting of microgel adsorption and non-Pickering stabilization. Moreover, the preparation of HIPEs provided the possibility of their application in porous materials and responsive materials.

Polymerizable Molecular Silsesquioxane Cage Armored Hybrid Microcapsules with In Situ Shell Functionalization.

We prepared core-shell polymer-silsesquioxane hybrid microcapsules from cage-like methacryloxypropyl silsesquioxanes (CMSQs) and styrene (St). The presence of CMSQ can moderately reduce the interfacial tension between St and water and help to emulsify the monomer prior to polymerization. Dynamic light scattering (DLS) and TEM analysis demonstrated that uniform core-shell latex particles were achieved. The polymer latex particles were subsequently transformed into well-defined hollow nanospheres by removing the polystyrene (PS) core with 1:1 ethanol/cyclohexane. High-resolution TEM and nitrogen adsorption-desorption analysis showed that the final nanospheres possessed hollow cavities and had porous shells; the pore size was approximately 2-3 nm. The nanospheres exhibited large surface areas (up to 486 m2 g-1 ) and preferential adsorption, and they demonstrated the highest reported methylene blue adsorption capacity (95.1 mg g-1 ). Moreover, the uniform distribution of the methacryloyl moiety on the hollow nanospheres endowed them with more potential properties. These results could provide a new benchmark for preparing hollow microspheres by a facile one-step template-free method for various applications.

A facile synthesis of octa(carboxyphenyl)silsesquioxane.

A facile method was reported for a robust octa(carboxyphenyl) functionalized silsesquioxane (OCS), which offers excellent potential as a nanobuilding-block for the preparation of functional materials. OCS was fully characterized by FT-IR, (1)H, (13)C, (29)Si-NMR, Maldi-TOF-MS spectra, and W-XRD analysis.

Structure-property relationship of highly π-conjugated Schiff-base moiety in liquid crystal diepoxide polymerization and mesophases stabilization.

A study of the structure-property relationship of the highly π-conjugated Schiff-base moiety in polymerization of a liquid crystal (LC) diepoxide oligomer (PBMBA) and mesophases stabilization has been investigated. We first proposed two exothermic peaks distinctly observed in nonisothermal polymerization curves for thermal copolymerization of PBMBA monomer with a diamine comonomer that corresponded to two different reactions, namely, an epoxy-amine polymerization and anionic polymerization. For PBMBA, note that an unexpected homopolymerization accompanying an appearance of enantiotropic mesophase transitions had taken place in the absence of any initiators, evidencing the possibility of an anionic mechanism yielding a homopolymer. A novel Schiff-base model compound (SBM) was synthesized and used to induce the polymerization of different types of epoxies. Based on the structure-property relationship, we considered a specific role of highly π-conjugated Schiff-base moieties in the anionic polymerization of PBMBA and hoped the mesophases could be stabilized using this mechanism, which may provide a key strategy for design of the polymer-dispersed liquid crystal (PDLC) materials via a chemical process.