HC2L: Hybrid and Cooperative Contrastive Learning for Cross-lingual Spoken Language Understanding.

State-of-the-art model for zero-shot cross-lingual spoken language understanding performs cross-lingual unsupervised contrastive learning to achieve the label-agnostic semantic alignment between each utterance and its code-switched data. However, it ignores the precious intent/slot labels, whose label information is promising to help capture the label-aware semantics structure and then leverage supervised contrastive learning to improve both source and target languages' semantics. In this paper, we propose Hybrid and Cooperative Contrastive Learning to address this problem. Apart from cross-lingual unsupervised contrastive learning, we design a holistic approach that exploits source language supervised contrastive learning, cross-lingual supervised contrastive learning and multilingual supervised contrastive learning to perform label-aware semantics alignments in a comprehensive manner. Each kind of supervised contrastive learning mechanism includes both single-task and joint-task scenarios. In our model, one contrastive learning mechanism's input is enhanced by others. Thus the total four contrastive learning mechanisms are cooperative to learn more consistent and discriminative representations in the virtuous cycle during the training process. Experiments show that our model obtains consistent improvements over 9 languages, achieving new state-of-the-art performance.

FXR/Menin-mediated epigenetic regulation of E2F3 expression controls ß-cell proliferation and is increased in islets from diabetic GK rats after RYGB.

Farnesoid X receptor (FXR) improves the function of islets, especially in the setting of Roux-en-Y gastric bypass (RYGB). Here we investigated how FXR activation regulates ß-cell proliferation and explored the potential link between FXR signaling and the menin pathway in controlling E2F3 expression, a key transcription factor for controlling adult ß-cell proliferation. Stimulation with the FXR agonist GW4064 or chenodeoxycholic acid (CDCA) increased E2F3 expression and ß-cell proliferation. Consistently, E2F3 knockdown abolished GW4064-induced proliferation. Treatment with GW4064 increased E2F3 expression in ß-cells via enhancing Steroid receptor coactivator-1 (SRC1) recruitment, increasing the pro-transcriptional acetylation of histone H3 at the E2f3 promoter. GW4064 treatment also decreased the association between FXR and menin, leading to the induction of FXR-mediated SRC1 recruitment. Mimicking the impact of FXR agonists, RYGB also increased E2F3 expression and ß-cell proliferation in GK rats and SD rats. These findings unravel the crucial role of the FXR/menin signaling in epigenetically controlling E2F3 expression and ß-cell proliferation, a mechanism possibly underlying RYGB-induced ß-cell proliferation.

Fish sonar image recognition algorithm based on improved YOLOv5.

Fish stock assessment is crucial for sustainable marine fisheries management in rangeland ecosystems. To address the challenges posed by the overfishing of offshore fish species and facilitate comprehensive deep-sea resource evaluation, this paper introduces an improved fish sonar image detection algorithm based on the you only look once algorithm, version 5 (YOLOv5). Sonar image noise often results in blurred targets and indistinct features, thereby reducing the precision of object detection. Thus, a C3N module is incorporated into the neck component, where depth-separable convolution and an inverse bottleneck layer structure are integrated to lessen feature information loss during downsampling and forward propagation. Furthermore, lowercase shallow feature layer is introduced in the network prediction layer to enhance feature extraction for pixels larger than $ 4 \times 4 $. Additionally, normalized weighted distance based on a Gaussian distribution is combined with Intersection over Union (IoU) during gradient descent to improve small target detection and mitigate the IoU's scale sensitivity. Finally, traditional non-maximum suppression (NMS) is replaced with soft-NMS, reducing missed detections due to occlusion and overlapping fish targets that are common in sonar datasets. Experiments show that the improved model surpasses the original model and YOLOv3 with gains in precision, recall and mean average precision of 2.3%, 4.7% and 2.7%, respectively, and 2.5%, 6.3% and 6.7%, respectively. These findings confirm the method's effectiveness in raising sonar image detection accuracy, which is consistent with model comparisons. Given Unmanned Underwater Vehicle advancements, this method holds the potential to support fish culture decision-making and facilitate fish stock resource assessment.

Co-Guiding for Multi-Intent Spoken Language Understanding.

Recent graph-based models for multi-intent SLU have obtained promising results through modeling the guidance from the prediction of intents to the decoding of slot filling. However, existing methods (1) only model the unidirectional guidance from intent to slot, while there are bidirectional inter-correlations between intent and slot; (2) adopt homogeneous graphs to model the interactions between the slot semantics nodes and intent label nodes, which limit the performance. In this paper, we propose a novel model termed Co-guiding Net, which implements a two-stage framework achieving the mutual guidances between the two tasks. In the first stage, the initial estimated labels of both tasks are produced, and then they are leveraged in the second stage to model the mutual guidances. Specifically, we propose two heterogeneous graph attention networks working on the proposed two heterogeneous semantics-label graphs, which effectively represent the relations among the semantics nodes and label nodes. Besides, we further propose Co-guiding-SCL Net, which exploits the single-task and dual-task semantics contrastive relations. For the first stage, we propose single-task supervised contrastive learning, and for the second stage, we propose co-guiding supervised contrastive learning, which considers the two tasks' mutual guidances in the contrastive learning procedure. Experiment results on multi-intent SLU show that our model outperforms existing models by a large margin, obtaining a relative improvement of 21.3% over the previous best model on MixATIS dataset in overall accuracy. We also evaluate our model on the zero-shot cross-lingual scenario and the results show that our model can relatively improve the state-of-the-art model by 33.5% on average in terms of overall accuracy for the total 9 languages.

DOA Estimation for Coherent Sources Based on Uniformly Distributed Two Concentric Rings Array.

The direction estimation of the coherent source in a uniform circular array is an essential part of the signal processing area of the array, but the traditional uniform circular array algorithm has a low localization accuracy and a poor localization effect on the coherent source. To solve this problem, this paper proposes a two-dimensional direction of arrival (DOA) estimation for the coherent source in broadband. Firstly, the central frequency of the coherent sound source is estimated using the frequency estimation method of the delayed data, and a real-valued beamformer is constructed using the concept of the multiloop phase mode. Then, the cost function in the beam space is obtained. Finally, the cost function is searched in two dimensions to locate the sound source. In this paper, we simulate the DOA of the sound source at different frequencies and signal-to-noise ratios and analyze the resolution of the circular array. The simulation results show that the proposed algorithm can estimate the direction of arrival with high precision and achieve the desired results.

RAB27B controls palmitoylation-dependent NRAS trafficking and signaling in myeloid leukemia.

RAS mutations are among the most prevalent oncogenic drivers in cancers. RAS proteins propagate signals only when associated with cellular membranes as a consequence of lipid modifications that impact their trafficking. Here, we discovered that RAB27B, a RAB family small GTPase, controlled NRAS palmitoylation and trafficking to the plasma membrane, a localization required for activation. Our proteomic studies revealed RAB27B upregulation in CBL- or JAK2-mutated myeloid malignancies, and its expression correlated with poor prognosis in acute myeloid leukemias (AMLs). RAB27B depletion inhibited the growth of CBL-deficient or NRAS-mutant cell lines. Strikingly, Rab27b deficiency in mice abrogated mutant but not WT NRAS-mediated progenitor cell growth, ERK signaling, and NRAS palmitoylation. Further, Rab27b deficiency significantly reduced myelomonocytic leukemia development in vivo. Mechanistically, RAB27B interacted with ZDHHC9, a palmitoyl acyltransferase that modifies NRAS. By regulating palmitoylation, RAB27B controlled c-RAF/MEK/ERK signaling and affected leukemia development. Importantly, RAB27B depletion in primary human AMLs inhibited oncogenic NRAS signaling and leukemic growth. We further revealed a significant correlation between RAB27B expression and sensitivity to MEK inhibitors in AMLs. Thus, our studies presented a link between RAB proteins and fundamental aspects of RAS posttranslational modification and trafficking, highlighting future therapeutic strategies for RAS-driven cancers.

Relational Temporal Graph Reasoning for Dual-Task Dialogue Language Understanding.

Dual-task dialog language understanding aims to tackle two correlative dialog language understanding tasks simultaneously via leveraging their inherent correlations. In this paper, we put forward a new framework, whose core is relational temporal graph reasoning. We propose a speaker-aware temporal graph (SATG) and a dual-task relational temporal graph (DRTG) to facilitate relational temporal modeling in dialog understanding and dual-task reasoning. Besides, different from previous works that only achieve implicit semantics-level interactions, we propose to model the explicit dependencies via integrating prediction-level interactions. To implement our framework, we first propose a novel model Dual-tAsk temporal Relational rEcurrent Reasoning network (DARER), which first generates the context-, speaker- and temporal-sensitive utterance representations through relational temporal modeling of SATG, then conducts recurrent dual-task relational temporal graph reasoning on DRTG, in which process the estimated label distributions act as key clues in prediction-level interactions. And the relational temporal modeling in DARER is achieved by relational graph convolutional networks (RGCNs). Then we further propose Relational Temporal Transformer (ReTeFormer), which achieves fine-grained relational temporal modeling via Relation- and Structure-aware Disentangled Multi-head Attention. Accordingly, we propose DARER with ReTeFormer (DARER 2), which adopts two variants of ReTeFormer to achieve the relational temporal modeling of SATG and DTRG, respectively. The extensive experiments on different scenarios verify that our models outperform state-of-the-art models by a large margin. Remarkably, on the dialog sentiment classification task in the Mastodon dataset, DARER and DARER 2 gain relative improvements of about 28% and 34% over the previous best model in terms of F1.

A 0.82 µVrms ultralow 1/f noise bandgap reference for a MEMS gyroscope.

High-precision microelectromechanical system (MEMS) gyroscopes are significant in many applications. Bias instability (BI) is an important parameter that indicates the performance of a MEMS gyroscope and is affected by the 1/f noise of the MEMS resonator and readout circuit. Since the bandgap reference (BGR) is an important block in the readout circuit, reducing its 1/f noise is key to improving a gyroscope's BI. In a traditional BGR, the error amplifier is applied to provide a virtual short-circuit point, but it introduces the main low-frequency noise sources. This paper proposes an ultralow 1/f noise BGR by removing the error amplifier and applying an optimized circuit topology. In addition, a simplified but accurate noise model of the proposed BGR is obtained to optimize the BGR's output noise performance. To verify this design, the proposed BGR has been implemented in a 180 nm CMOS process with a chip area of 545 × 423 µm. The experimental results show that the BGR's output integrated noise from 0.1 to 10 Hz is 0.82 µV and the thermal noise is 35 nV/√Hz. Furthermore, bias stability tests of the MEMS gyroscope fabricated in our laboratory with the proposed BGR and some commercial BGRs are carried out. Statistical results show that reducing the BGR's 1/f noise can nearly linearly improve the gyroscope's BI.

A swallowable X-ray dosimeter for the real-time monitoring of radiotherapy.

Monitoring X-ray radiation in the gastrointestinal tract can enhance the precision of radiotherapy in patients with gastrointestinal cancer. Here we report the design and performance, in the gastrointestinal tract of rabbits, of a swallowable X-ray dosimeter for the simultaneous real-time monitoring of absolute absorbed radiation dose and of changes in pH and temperature. The dosimeter consists of a biocompatible optoelectronic capsule containing an optical fibre, lanthanide-doped persistent nanoscintillators, a pH-sensitive polyaniline film and a miniaturized system for the wireless readout of luminescence. The persistent luminescence of the nanoscintillators after irradiation can be used to continuously monitor pH without the need for external excitation. By using a neural-network-based regression model, we estimated the radiation dose from radioluminescence and afterglow intensity and temperature, and show that the dosimeter was approximately five times more accurate than standard methods for dose determination. Swallowable dosimeters may help to improve radiotherapy and to understand how radiotherapy affects tumour pH and temperature.

Exploration of the intelligent-auxiliary design of architectural space using artificial intelligence model.

In order to carry out a comprehensive design description of the specific architectural model of AI, the auxiliary model of AI and architectural spatial intelligence is deeply integrated, and flexible design is carried out according to the actual situation. AI assists in the generation of architectural intention and architectural form, mainly supporting academic and working theoretical models, promoting technological innovation, and thus improving the design efficiency of the architectural design industry. AI-aided architectural design enables every designer to achieve design freedom. At the same time, with the help of AI, architectural design can complete the corresponding work faster and more efficiently. With the help of AI technology, through the adjustment and optimization of keywords, AI automatically generates a batch of architectural space design schemes. Against this background, the auxiliary model of architectural space design is established through the literature research of the AI model, the architectural space intelligent auxiliary model, and the semantic network and the internal structure analysis of architectural space. Secondly, to ensure compliance with the three-dimensional characteristics of the architectural space from the data source, based on the analysis of the overall function and structure of space design, the intelligent design of the architectural space auxiliary by Deep Learning is carried out. Finally, it takes the 3D model selected in the UrbanScene3D data set as the research object, and the auxiliary performance of AI's architectural space intelligent model is tested. The research results show that with the increasing number of network nodes, the model fitting degree on the test data set and training data set is decreasing. The fitting curve of the comprehensive model shows that the intelligent design scheme of architectural space based on AI is superior to the traditional architectural design scheme. As the number of nodes in the network connection layer increases, the intelligent score of space temperature and humidity will continue to rise. The model can achieve the optimal intelligent auxiliary effect of architectural space. The research has practical application value for promoting the intelligent and digital transformation of architectural space design.

Sea Cucumber Detection Algorithm Based on Deep Learning.

The traditional single-shot multiBox detector (SSD) for the recognition process in sea cucumbers has problems, such as an insufficient expression of features, heavy computation, and difficulty in application to embedded platforms. To solve these problems, we proposed an improved algorithm for sea cucumber detection based on the traditional SSD algorithm. MobileNetv1 is selected as the backbone of the SSD algorithm. We increase the feature receptive field by receptive field block (RFB) to increase feature details and location information of small targets. Combined with the attention mechanism, features at different depths are strengthened and irrelevant features are suppressed. The experimental results show that the improved algorithm has better performance than the traditional SSD algorithm. The average precision of the improved algorithm is increased by 5.1%. The improved algorithm is also more robust. Compared with YOLOv4 and the Faster R-CNN algorithm, the performance of this algorithm on the P-R curve is better, indicating that the performance of this algorithm is better. Thus, the improved algorithm can stably detect sea cucumbers in real time and provide reliable feedback information.

Potent suppression of neuroendocrine tumors and gastrointestinal cancers by CDH17CAR T cells without toxicity to normal tissues.

Gastrointestinal cancers (GICs) and neuroendocrine tumors (NETs) are often refractory to therapy after metastasis. Adoptive cell therapy using chimeric antigen receptor (CAR) T cells, though remarkably efficacious for treating leukemia, is yet to be developed for solid tumors such as GICs and NETs. Here we isolated a llama-derived nanobody, VHH1, and found that it bound cell surface adhesion protein CDH17 upregulated in GICs and NETs. VHH1-CAR T cells (CDH17CARTs) killed both human and mouse tumor cells in a CDH17-dependent manner. CDH17CARTs eradicated CDH17-expressing NETs and gastric, pancreatic and colorectal cancers in either tumor xenograft or autochthonous mouse models. Notably, CDH17CARTs do not attack normal intestinal epithelial cells, which also express CDH17, to cause toxicity, likely because CDH17 is localized only at the tight junction between normal intestinal epithelial cells. Thus, CDH17 represents a class of previously unappreciated tumor-associated antigens that is 'masked' in healthy tissues from attack by CAR T cells for developing safer cancer immunotherapy.

Prolactin-regulated Pbk is involved in pregnancy-induced ß-cell proliferation in mice.

Gestational diabetes mellitus (GDM) is a condition of diabetes with onset or first recognition in pregnancy. Its incidence is increasing, and GDM deleteriously affects both mother and the fetus during and even after pregnancy. Previous studies in mice have shown that during pregnancy, ß-cell proliferation increases in the middle and late stages of pregnancy and returns to normal levels after delivery. Hormones, such as prolactin, estradiol, and progesterone as well as protein kinases, play important roles in regulating gestation-mediated ß-cell proliferation; however, the regulatory relationship between them is uncertain. We previously found that protein kinase Pbk was crucial for basal proliferation of mouse islet cells. Herein we show that Pbk is upregulated during pregnancy in mice and Pbk kinase activity is required for enhanced ß- cell proliferation during pregnancy. Notably, knock-in (KI) of a kinase-inactivating Pbk mutation leads to impaired glucose tolerance and reduction of ß-cell proliferation and islet mass in mice during pregnancy. Prolactin upregulates the expression of Pbk, but the upregulation is diminished by knockdown of the prolactin receptor and by the inhibitors of JAK and STAT5, which mediate prolactin receptor signaling, in ß-cells. Treatment of ß-cells with prolactin increases STAT5 binding to the Pbk locus, as well as the recruitment of RNA polymerase II, resulting in increased Pbk transcription. These results demonstrate that Pbk is upregulated during pregnancy, at least partly by prolactin-induced and STAT5-mediated enhancement of gene transcription, and Pbk is essential for pregnancy-induced ß-cell proliferation, increase in islet mass, and maintenance of normal blood glucose during pregnancy in preclinical models. These findings provide new insights into the interplay between hormones and protein kinases that ultimately prevent the development of GDM.

Menin-regulated Pbk controls high fat diet-induced compensatory beta cell proliferation.

Pancreatic beta cells undergo compensatory proliferation in the early phase of type 2 diabetes. While pathways such as FoxM1 are involved in regulating compensatory beta cell proliferation, given the lack of therapeutics effectively targeting beta cell proliferation, other targetable pathways need to be identified. Herein, we show that Pbk, a serine/threonine protein kinase, is essential for high fat diet (HFD)-induced beta cell proliferation in vivo using a Pbk kinase deficiency knock-in mouse model. Mechanistically, JunD recruits menin and HDAC3 complex to the Pbk promoter to reduce histone H3 acetylation, leading to epigenetic repression of Pbk expression. Moreover, menin inhibitor (MI) disrupts the menin-JunD interaction and augments Pbk transcription. Importantly, MI administration increases beta cell proliferation, ameliorating hyperglycemia, and impaired glucose tolerance (IGT) in HFD-induced diabetic mice. Notably, Pbk is required for the MI-induced beta cell proliferation and improvement of IGT. Together, these results demonstrate the repressive role of the menin/JunD/Pbk axis in regulating HFD-induced compensatory beta cell proliferation and pharmacologically regulating this axis may serve as a novel strategy for type 2 diabetes therapy.

Wild-type menin is rapidly degraded via the ubiquitin-proteasome pathway in a rat insulinoma cell line.

Menin is encoded by multiple endocrine neoplasia type 1 (MEN1) gene, the germ line mutations of which are the main cause of pancreatic neuroendocrine tumors (PNETs). To date, a large number of frameshift, nonsense and missense mutations of MEN1 have been identified to be responsible for part of MEN1-defficient PNETs patients due to truncation or rapid degradation of menin protein. However, the stability of the wild-type (WT) menin in PNETs is totally unknown. In the present study, we observed ubiquitination of WT menin in 293T cells by transfection of ectopic WT menin and HA-ubiquitin. As expected, either endogenous or ectopic WT menin is stable in 293T cells, whereas in INS-1 cells, a rat insulinoma cell line derived from PNETs, either endogenous or ectopic WT menin is rapidly degraded through ubiquitin-proteasome pathway. Furthermore, the degradation of WT menin is more rapid in the presence of serum. Our findings suggest that in part of PNETs patients with WT MEN1, a ubiquitin-proteasome system targeting menin is untimely activated.

Periodic Nonlinear Error Analysis and Compensation of a Single-Excited Petal-Shaped Capacitive Encoder to Achieve High-Accuracy Measurement.

The measurement results of a single-excitation petal-shaped capacitive encoder show strong periodic characteristics for nonlinear errors. This paper presents the analysis of periodic nonlinear errors in a single-excitation petal-shaped encoder in terms of three main aspects-sensitive structure processing error, circuit demodulation error, and installation error. Analytical and simulation results confirm that the first-, second-, and fourth-periodic electrical errors are caused by the misalignment of circuit parameters, non-uniform segmentation of the processing error, and cross interference of the electric field, respectively. Further experimental investigation reveals that the mechanical periodic error is caused by installation misalignment. Based on these analytical, simulation, and experimental results, the design of the capacitive encoder was optimized and a method based on harmonic components was applied to compensate the periodic nonlinear error of the encoder. Measurement results shows that the prototype which has 180 petal-shaped numbers can achieve a reduction of periodic nonlinear errors to less than 0.02° and its accuracy can be improved to 0.0006° after compensation over the full measurement range.

GLP-1 signaling suppresses menin's transcriptional block by phosphorylation in ß cells.

Both menin and glucagon-like peptide 1 (GLP-1) pathways play central yet opposing role in regulating ß cell function, with menin suppressing, and GLP-1 promoting, ß cell function. However, little is known as to whether or how GLP-1 pathway represses menin function. Here, we show that GLP-1 signaling-activated protein kinase A (PKA) directly phosphorylates menin at the serine 487 residue, relieving menin-mediated suppression of insulin expression and cell proliferation. Mechanistically, Ser487-phosphorylated menin gains increased binding affinity to nuclear actin/myosin IIa proteins and gets sequestrated from the Ins1 promoter. This event leads to reduced binding of repressive epigenetic histone modifiers suppressor variegation 3-9 homologue protein 1 (SUV39H1) and histone deacetylases 1 (HDAC1) at the locus and subsequently increased Ins1 gene transcription. Ser487 phosphorylation of menin also increases expression of proproliferative cyclin D2 and ß cell proliferation. Our results have uncovered a previously unappreciated physiological link in which GLP-1 signaling suppresses menin function through phosphorylation-triggered and actin/myosin cytoskeletal protein-mediated derepression of gene transcription.

Menin Upregulates FOXO1 Protein Stability by Repressing Skp2-Mediated Degradation in ß Cells.

OBJECTIVES: Menin, a chromatin binding protein, interacts with various epigenetic regulators to regulate gene transcription, whereas forkhead box protein O1 (FOXO1) is a transcription factor that can be regulated by multiple signaling pathways. Both menin and FOXO1 are crucial regulators of ß-cell function and metabolism; however, whether or how they interplay to regulate ß cells is not clear. METHODS: To examine whether menin affects expression of FOXO1, we ectopically expressed menin complementary DNA and small hairpin RNA targeting menin via a retroviral vector in INS-1 cells. Western blotting was used to analyze protein levels. RESULTS: Our current work shows that menin increases the expression of FOXO1. Menin stabilizes FOXO1 protein level in INS-1 cells, as shown by increased half-life of FOXO1 by menin expression. Moreover, menin represses ubiquitination of FOXO1 protein and AKT phosphorylation, We found that menin stabilizes FOXO1 by repressing FOXO1 degradation mediated by S-phase kinase-associated protein 2 (Skp2), an E3 ubiquitin ligase, promoting caspase 3 activation and apoptosis. CONCLUSIONS: Because FOXO1 upregulates the menin gene transcription, our findings unravel a crucial menin and FOXO1 interplay, with menin and FOXO1 upregulating their expression reciprocally, forming a positive feedback loop to sustain menin and FOXO1 expression.