arXiv subjects

Takashi Miyazaki

Publications and source records attributed to Takashi Miyazaki.

5 recordsLinked to original sources

arXiv2026

Findings of sub-$T_\mathrm{g}$ endotherm in vapor-deposited ultrastable phenolphthalein glass

We have performed heat capacity measurements for physically vapor-deposited phenolphthalein glass using conventional and in-house high-sensitivity differential scanning calorimetry. As a result, we found that phenolphthalein forms an ultrastable glass when deposited at 313 K, about 0.86 times the ordinary glass transition temperature of 361 K. In addition, we observed a large endotherm (sub-$T_\mathrm{g}$ endotherm) between deposition and the ordinary glass transition temperature. From an enthalpy perspective, the integrated enthalpy of the sub-$T_\mathrm{g}$ endotherm increased with decreasing film thickness following an inverse-power dependence, whose exponent increased as the deposition rate decreased. In particular, the integrated enthalpy of the sub-$T_\mathrm{g}$ endotherm was comparable in magnitude to the enthalpy of fusion of crystalline phenolphthalein. To investigate the structural origin of the sub-$T_\mathrm{g}$ endotherm, we performed wide-angle X-ray diffraction and in situ atomic force microscopy. As a result, we found that while the stable structure is linked to an anisotropic amorphous structure consistent with past research on ultrastable glasses, the appearance of the sub-$T_\mathrm{g}$ endotherm is associated with surface morphological relaxation, as suggested by atomic force microscopy.

cond-mat.soft↗

arXiv2022

Ladder Siamese Network: a Method and Insights for Multi-level Self-Supervised Learning

Siamese-network-based self-supervised learning (SSL) suffers from slow convergence and instability in training. To alleviate this, we propose a framework to exploit intermediate self-supervisions in each stage of deep nets, called the Ladder Siamese Network. Our self-supervised losses encourage the intermediate layers to be consistent with different data augmentations to single samples, which facilitates training progress and enhances the discriminative ability of the intermediate layers themselves. While some existing work has already utilized multi-level self supervisions in SSL, ours is different in that 1) we reveal its usefulness with non-contrastive Siamese frameworks in both theoretical and empirical viewpoints, and 2) ours improves image-level classification, instance-level detection, and pixel-level segmentation simultaneously. Experiments show that the proposed framework can improve BYOL baselines by 1.0% points in ImageNet linear classification, 1.2% points in COCO detection, and 3.1% points in PASCAL VOC segmentation. In comparison with the state-of-the-art methods, our Ladder-based model achieves competitive and balanced performances in all tested benchmarks without causing large degradation in one.

cs.CV↗

arXiv2018

Deep Learning Based Multi-modal Addressee Recognition in Visual Scenes with Utterances

With the widespread use of intelligent systems, such as smart speakers, addressee recognition has become a concern in human-computer interaction, as more and more people expect such systems to understand complicated social scenes, including those outdoors, in cafeterias, and hospitals. Because previous studies typically focused only on pre-specified tasks with limited conversational situations such as controlling smart homes, we created a mock dataset called Addressee Recognition in Visual Scenes with Utterances (ARVSU) that contains a vast body of image variations in visual scenes with an annotated utterance and a corresponding addressee for each scenario. We also propose a multi-modal deep-learning-based model that takes different human cues, specifically eye gazes and transcripts of an utterance corpus, into account to predict the conversational addressee from a specific speaker's view in various real-life conversational scenarios. To the best of our knowledge, we are the first to introduce an end-to-end deep learning model that combines vision and transcripts of utterance for addressee recognition. As a result, our study suggests that future addressee recognition can reach the ability to understand human intention in many social situations previously unexplored, and our modality dataset is a first step in promoting research in this field.

cs.AI↗

arXiv2010

A modified porous titanium sheet prepared by plasma activated sintering for biomedical applications

This study aimed to develop a contamination free porous titanium scaffold by a plasma activated sintering within an originally developed TiN coated graphite mold. The surface of porous titanium sheet with or without a coated graphite mold was characterized. The cell adhesion property of porous titanium sheet was also evaluated in this study. The peak of TiC was detected on the titanium sheet processed with the graphite mold without a TiN coating. Since the titanium fiber elements were directly in contact with the carbon graphite mold during processing, surface contamination was unavoidable event in this condition. The TiC peak was not detectable on the titanium sheet processed within the TiN coated carbon graphite mold. This modified plasma activated sintering with the TiN coated graphite mold would be useful to fabricate a contamination free titanium sheet. The number of adherent cells on the modified titanium sheet was greater than that of the bare titanium plate. Stress fiber formation and the extension of the cells were observed on the titanium sheets. This modified titanium sheet is expected to be a new tissue engineering material in orthopedic bone repair.

q-bio.CB↗

arXiv2010

Bone regenerative potential of mesenchymal stem cells on a micro-structured titanium processed by wire-type electric discharge machining

A new strategy with bone tissue engineering by mesenchymal stem cell transplantation on titanium implant has been dawn attention. The surface scaffold properties of titanium surface play an important role in bone regenerative potential of cells. The surface topography and chemistry are postulated to be two major factors increasing the scaffold properties of titanium implants. This study aimed to evaluate the osteogenic gene expression of mesenchymal stem cells on titanium processed by wire-type electric discharge machining. Some amount of roughness and distinctive irregular features were observed on titanium processed by wire-type electric discharge machining. The thickness of suboxide layer was concomitantly grown during the processing. Since the thickness of oxide film and micro-topography allowed an improvement of mRNA expression of cells, titanium processed by wire-type electric discharge machining is a promising candidate for mesenchymal stem cell based functional restoration of implants.

q-bio.CB↗