Ichiro Takeuchi

A new species of Caprella Lamarck, 1801 (Crustacea: Amphipoda: Caprellidae) collected from the rocky coasts of Ehime Prefecture, western Japan, is described. The sampling site of the present new species is adjacent to the district where organic citrus cultivation is predominant. Of the approximately 200 species of Caprella, the present species, C. gracilenta sp. nov., is most closely related to C. rhopalochir Mayer, 1890. The type locality of C. rhopalochir is the mouth of the Amur River, located at the northern end of the Sea of Japan. In addition, C. rhopalochir was recorded until the 1970s from northern Kyushu and adjacent areas near the southwestern end of the Sea of Japan, where oceanographic conditions are largely different from the type locality of C. rhopalochir. Caprella gracilenta sp. nov. is distinguished from C. rhopalochir by the absence of forwardly curved triangular projection near the posterior end of pereonite 1, triangular projection on the dorsal face of the propodus of gnathopod 2, paired dorsal projections on pereonites 3–4, and triangular projection near the posterior end of pereonites 3–4. Of the two descriptions of C. rhopalochir from near the southwestern end of the Sea of Japan, a description from Saga Prefecture was identified as C. gracilenta sp. nov. This re-identification indicates that C. gracilenta sp. nov. is a rare species widely distributed along the western coast of Japan.

Channelrhodopsins (ChRs) are light-gated ion channels. These proteins are widely used in optogenetics to optically manipulate neural activity. However, manipulation using short-wavelength light to activate ChRs causes cell toxicity and is hampered by low tissue penetration. To overcome these difficulties, although several red-shifted ChR variants have been identified, further red-shift is required for more efficient and noninvasive neural control. While molecular screening of ChRs requires high-cost experiments, recent machine-learning-based protein functionality prediction enables more efficient selection of target proteins for characterization. Here, we constructed an elastic-net machine-learning model trained on 1,163 experimental data to predict the maximum absorption wavelength ( λ _max ) of uncharacterized ChRs. The model suggested several red-shifted candidates, and we identified a viral rhodopsin channel, ChR024, with the second-longest λ _max as a cation-conducting ChR ( λ _max = ∼578 nm) after Chrimson ( λ _max = ∼580 nm). This result demonstrates the high impact of ML on reducing the screening costs of functional proteins.

Irgarol 1051 and Diuron are used as boosters in copper-based anti-fouling paints, as well as in agriculture and home gardening. We monitored the coral reefs and adjacent waters of northern Okinawa Island three times, i.e., in August 2015, June 2016 and December 2016. The mean concentrations were 1.44 ng/L for Irgarol and 1.07 ng/L for Diuron. Although the range of these concentrations was lower than those previously recorded in southern Japan, Irgarol and Diuron were significantly higher in December 2016 than in June 2016. Several estimations of Predicted No Effect Concentrations (PNECs) have recently been published. In December 2016, compared to the lowest PNEC concentration of each substances, the concentrations of Irgarol and Diuron exceeded the PNEC at nine sites and three sites, respectively. These results suggest that the effects of low concentrations of Irgarol and Diuron, close to the PNEC, on hermatypic corals need to be investigated. The regulation of anti-fouling paints in Japan is self-regulated through industry cooperation. Irgarol and Diuron have not been registered as anti-fouling paints for fishing nets by the fishing cooperatives. The significant increase of these substances suggests that the activity of vessels with paints containing Irgarol has increased in December 2016. This study highlights the need for continuous monitoring of these herbicides in coral reefs and adjacent waters.

In this paper, we introduce si4onnx, a package for performing selective inference on deep learning models. Techniques such as CAM in XAI and reconstruction-based anomaly detection using VAE can be interpreted as methods for identifying significant regions within input images. However, the identified regions may not always carry meaningful significance. Therefore, evaluating the statistical significance of these regions represents a crucial challenge in establishing the reliability of AI systems. si4onnx is a Python package that enables straightforward implementation of hypothesis testing with controlled type I error rates through selective inference. It is compatible with deep learning models constructed using common frameworks such as PyTorch and TensorFlow.

The 11 March 2011 Great East Japan Earthquake generated a massive tsunami that triggered an extraordinary release of floating transoceanic marine debris. That floating debris subsequently landed on coastlines from Midway Atoll to the Hawaiʻian Islands and from south central Alaska to central California of North America. Two species of Caprella (Crustacea: Amphipoda: Caprellidae) were collected from a large concrete fisheries dock (approximately 20 m length) that was part of the floating debris, on Agate Beach, near Newport, Oregon, USA, on 5 June 2012. The caprellids recovered from the dock are identified as Caprella cristibrachium Mayer, 1903 and C. mutica Schurin, 1935. Caprella cristibrachium had not been observed previously outside of the Bering Sea to the Far East Asia. The robust body somites and short antennae and basis of gnathopod 2 make C. cristibrachium suitable for attaching the filamentous macroalgae and/or marine invertebrates on the large tsunami fishing dock. Caprella mutica, originally recorded from the Far East Asia, was previously introduced to Oregon, to other cold-water areas between northern California and Alaska, the eastern and western North Atlantic and to the southern hemisphere.

Latent Space Bayesian Optimization (LSBO) combines generative models, typically Variational Autoencoders (VAE), with Bayesian Optimization (BO) to generate de-novo objects of interest. However, LSBO faces challenges due to the mismatch between the objectives of BO and VAE, resulting in poor exploration capabilities. In this paper, we propose novel contributions to enhance LSBO efficiency and overcome this challenge. We first introduce the concept of latent consistency/inconsistency as a crucial problem in LSBO, arising from the VAE-BO mismatch. To address this, we propose the Latent Consistent Aware-Acquisition Function (LCA-AF) that leverages consistent points in LSBO. Additionally, we present LCA-VAE, a novel VAE method that creates a latent space with increased consistent points through data augmentation in latent space and penalization of latent inconsistencies. Combining LCA-VAE and LCA-AF, we develop LCA-LSBO. Our approach achieves high sample-efficiency and effective exploration, emphasizing the significance of addressing latent consistency through the novel incorporation of data augmentation in latent space within LCA-VAE in LSBO. We showcase the performance of our proposal via de-novo image generation and de-novo chemical design tasks.

In this study, we propose a method Distributionally Robust Safe Screening (DRSS), for identifying unnecessary samples and features within a DR covariate shift setting. This method effectively combines DR learning, a paradigm aimed at enhancing model robustness against variations in data distribution, with safe screening (SS), a sparse optimization technique designed to identify irrelevant samples and features prior to model training. The core concept of the DRSS method involves reformulating the DR covariate-shift problem as a weighted empirical risk minimization problem, where the weights are subject to uncertainty within a predetermined range. By extending the SS technique to accommodate this weight uncertainty, the DRSS method is capable of reliably identifying unnecessary samples and features under any future distribution within a specified range. We provide a theoretical guarantee of the DRSS method and validate its performance through numerical experiments on both synthetic and real-world datasets.