Smart-building load forecasters are often trained offline on dense, multivariate, high-frequency data, but deployment may provide only hourly, feature-limited inputs. Missing features must then be reconstructed, and their errors can propagate through the model. If this input uncertainty is not reflected, prediction intervals may become miscalibrated, affecting demand-response scheduling. Our work examines where uncertainty should be placed once inference inputs are reconstructed. We develop a unified one-day-ahead probabilistic forecasting framework that aligns temporal resolution, reconstructs the unavailable inputs, and derives causal features, and we compare a modular post-hoc residual-quantile scheme with an integrated in-model quantile-learning scheme. The comparison uses three mid-scale Deep Learning (DL) backbones: recurrent, hybrid recurrent, and attention-based Temporal Fusion Transformer (TFT) models, under identical inputs, forecasting horizon, preprocessing rules, and training budgets. Results show that uncertainty placement is backbone-dependent. Integrated quantile learning is most reliable with the TFT, yielding 2.2-3.6% MAPE and 28-83 W RMSE on the labeled test window, while producing intervals about 5 & times; narrower than the modular intervals at the closest-to-nominal coverage level. Diebold-Mariano tests support the TFT ranking and the mixed behavior of the recurrent backbones. A reconstruction-sensitivity test shows that reconstructed inputs increase the Quantile Score (QS) by 106% while interval width remains nearly unchanged, indicating that the model does not automatically absorb reconstruction-induced uncertainty. Robustness checks against non-DL baselines and seasonal hold-out weeks support this ranking. Our results expose the limits of post-hoc residual quantiles when inference depends on reconstructed inputs.
Carbon fibers (CFs) with high tensile strength are highly desirable as reinforcement materials in advanced composites. However, their theoretical tensile strength (similar to 180 GPa), estimated from sp2 C-C bond strength in defect-free structures, remains unattainable because of intrinsic surface and internal defects. Here, we demonstrate an atomic layer deposition (ALD)-based strategy enabling dual-scale defect mitigation in commercial polyacrylonitrile-based CFs through simultaneous surface passivation and internal structural reinforcement. Conformal Al2O3 layers reduced surface irregularities and stress concentration sites, while infiltrated Al-containing species promoted interfacial bonding and cross-linking within the carbon microstructure. Consequently, the tensile strength of commercial CFs was significantly enhanced after ALD treatment, with T1000G fibers exhibiting a similar to 41% increase in tensile strength and reaching 8.45 +/- 1.35 GPa. Structural analyses using X-ray diffraction, Raman spectroscopy, small-angle X-ray scattering, field-emission transmission electron microscopy, and energy-dispersive X-ray spectroscopy mapping revealed reduced microvoid dimensions without significant alteration of the graphitic microstructure. These findings establish ALD-induced dual-scale defect mitigation as an effective and potentially industrially applicable approach for producing next-generation ultra-strong CFs. (c) 2026 Published by Elsevier Ltd on behalf of The editorial office of Journal of Materials Science & Technology. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Claudins (CLDNs) are a family of integral membrane proteins central to the formation of tight junctions, structures that are involved in paracellular transport and cellular growth and differentiation, and are critical for the maintenance of cellular polarity. Recent studies have provided evidence that CLDNs are aberrantly expressed in diverse types of human cancers, including hepatocellular carcinomas (HCCs). However, little is known about how CLDN expression is involved in cancer progression. In this study, we show that CLDN1 has a causal role in the epithelial–mesenchymal transition (EMT) in human liver cells, and that the c-Abl-Ras-Raf-1-ERK1/2 signaling axis is critical for the induction of malignant progression by CLDN1. Overexpression of CLDN1 induced expression of the EMT-regulating transcription factors Slug and Zeb1, and thereby led to repression of E-cadherin, β-catenin expression, enhanced expression of N-cadherin and Vimentin, a loss of cell adhesion, and increased cell motility in normal liver cells and HCC cells. In line with these findings, inhibition of either c-Abl or ERK clearly attenuated CLDN1-induced EMT, as evidenced by a reversal of N-cadherin and E-cadherin expression patterns, and restored normal motility. Collectively, these results indicate that CLDN1 is necessary for the induction of EMT in human liver cells, and that activation of the c-Abl-Ras-Raf-1-ERK1/2 signaling pathway is required for CLDN1-induced acquisition of the malignant phenotype. The present observations suggest that CLDN1 could be exploited as a biomarker for liver cancer metastasis and might provide a pivotal point for therapeutic intervention in HCC.
Impaired autophagy has been implicated in many neurodegenerative diseases, such as Parkinson's disease (PD), and might be responsible for deposition of aggregated proteins in neurons. However, little is known about how neuronal autophagy and clearance of aggregated proteins are regulated. Here, we show a role for Toll-like receptor 2 (TLR2), a pathogen-recognizing receptor in innate immunity, in regulation of neuronal autophagy and clearance of α-synuclein, a protein aggregated in synucleinopathies, including in PD. Activation of TLR2 resulted in the accumulation of α-synuclein aggregates in neurons as a result of inhibition of autophagic activity through regulation of the AKT/mTOR pathway. In contrast, inactivation of TLR2 resulted in autophagy activation and increased clearance of neuronal α-synuclein, and hence reduced neurodegeneration, in transgenic mice and in in vitro models. These results uncover roles of TLR2 in regulating neuronal autophagy and suggest that the TLR2 pathway may be targeted for autophagy activation strategies in treating neurodegenerative disorders.
