We present a solution for intelligent posture training based on accurate, real-time sitting posture monitoring using the LifeChair IoT cushion and supervised machine learning from pressure sensing and user body data. We demonstrate our system’s performance in sitting posture and seated stretch recognition tasks with over 98.82% accuracy in recognizing 15 different sitting postures and 97.94% in recognizing six seated stretches. We also show that user BMI divergence significantly affects posture recognition accuracy using machine learning. We validate our method’s performance in five different real-world workplace environments and discuss training strategies for the machine learning models. Finally, we propose the first smart posture data-driven stretch recommendation system in alignment with physiotherapy standards.

We present a new catalog of 9318 Lyα emitter (LAE) candidates at z = 2.2, 3.3, 4.9, 5.7, 6.6, and 7.0 that are photometrically selected by the SILVERRUSH program with a machine learning technique from large area (up to 25.0 deg²) imaging data with six narrowband filters taken by the Subaru Strategic Program with Hyper Suprime-Cam and a Subaru intensive program, Cosmic HydrOgen Reionization Unveiled with Subaru. We construct a convolutional neural network that distinguishes between real LAEs and contaminants with a completeness of 94% and a contamination rate of 1%, enabling us to efficiently remove contaminants from the photometrically selected LAE candidates. We confirm that our LAE catalogs include 177 LAEs that have been spectroscopically identified in our SILVERRUSH programs and previous studies, ensuring the validity of our machine learning selection. In addition, we find that the object-matching rates between our LAE catalogs and our previous results are ≃80%–100% at bright NB magnitudes of ≲24 mag. We also confirm that the surface number densities of our LAE candidates are consistent with previous results. Our LAE catalogs will be made public on our project webpage.

To estimate the functions of mitochondria of diverse eukaryotic nonmodel organisms in which the mitochondrial proteomes are not available, it is necessary to predict the protein sequence features of the mitochondrial proteins computationally. Various prediction methods that are trained using the proteins of model organisms belonging particularly to animals, plants, and fungi exist. However, such methods may not be suitable for predicting the proteins derived from nonmodel organisms because the sequence features of the mitochondrial proteins of diversified nonmodel organisms can differ from those of model organisms that are present only in restricted parts of the tree of eukaryotes. Here, we proposed NommPred, which predicts the mitochondrial proteins of nonmodel organisms that are widely distributed over eukaryotes. We used a gradient boosting machine to develop 2 predictors—one for predicting the proteins of mitochondria and the other for predicting the proteins of mitochondrion-related organelles that are highly reduced mitochondria. The performance of both predictors was found to be better than that of the best method available.

Entity Matching (EM) involves identifying different data representations referring to the same entity from multiple data sources and is typically formulated as a binary classification problem. It is a challenging problem in data integration due to the heterogeneity of data representations. State-of-the-art solutions have adopted NLP techniques based on pre-trained language models (PrLMs) via the fine-tuning paradigm, however, sequential fine-tuning of overparameterized PrLMs can lead to catastrophic forgetting, especially in low-resource scenarios. In this study, we propose a parameter-efficient paradigm for fine-tuning PrLMs based on adapters, small neural networks encapsulated between layers of a PrLM, by optimizing only the adapter and classifier weights while the PrLMs parameters are frozen. Adapter-based methods have been successfully applied to multilingual speech problems achieving promising results, however, the effectiveness of these methods when applied to EM is not yet well understood, particularly for generalized EM with heterogeneous data. Furthermore, we explore using (i) pre-trained adapters and (ii) invertible adapters to capture token-level language representations and demonstrate their benefits for transfer learning on the generalized EM benchmark. Our results show that our solution achieves comparable or superior performance to full-scale PrLM fine-tuning and prompt-tuning baselines while utilizing a significantly smaller computational footprint of the PrLM parameters.