A squaramide catalyzed regiospecific and stereoselective [4+2] cyclization of 2-benzothiazolimines with azlactones has been established.

This paper proposes DoubleDiffusion, a novel framework that combines heat dissipation diffusion and denoising for direct generative learning on 3D mesh surfaces. Our approach addresses the challenges of generating continuous signal distributions residing curve manifold surface. Unlike previous methods rely unrolling meshes into 2D or adopting field representations, DoubleDiffusion leverages Laplacian-Beltrami operator to process features respecting structure. combination enables effective...

An asymmetric [3+3] cyclization of nitroenynes and 3-pyrrolyloxindoles has been realized with a chiral bifunctional squaramide catalyst. This Michael/Friedel–Crafts cascade strategy provides facile efficient access to enantioenriched polycyclic aza-spirooxindoles 32–95% isolated yields excellent stereocontrol under mild reaction conditions.

We describe here an organocatalytic asymmetric cascade formal [3 + 3] cycloaddition of benzothiazoles with 2-nitroallylic acetates and nitroenynes. This dearomative methodology provided a facile efficient strategy for the construction broad range valuable benzothiazolopyridines bearing two adjacent stereogenic centers in moderate to good yields excellent stereocontrol.

The introduction of deep neural networks (DNNs) leads to a significant improvement the automatic speech recognition (ASR) performance. However, whole ASR system remains sophisticated due dependent on hidden Markov model (HMM). Recently, new end-to-end framework, which utilizes recurrent (RNNs) directly context-independent targets with connectionist temporal classification (CTC) objective function, is proposed and achieves comparable results hybrid HMM/DNN system. In this paper, we...

In this paper, we propose a deep convolutional neural network-based acoustic word embedding system for code-switching query by example spoken term detection. Different from previous configurations, combine audio data in two languages training instead of only using one single language. We trans-form the features keyword templates and searching content segments obtained sliding manner to fixed-dimensional vectors calculate distances between them. An auxiliary variability-invariant loss is also...

A new acoustic model based on deep neural network (DNN) has been introduced recently and outperforms the conventional Gaussian mixture (GMM) in speech recognition several tasks. However, number of parameters required by a DNN is much larger than that its counterpart. The excessive cost computation cumbers implementation many scenarios. In this paper, DNN-based recognizer implemented an embedded platform. To reduce size cost, converted from float-point to fixed-point NEON instructions are...

In this paper, we propose a deep convolutional neural network-based acoustic word embedding system on code-switching query by example spoken term detection. Different from previous configurations, combine audio data in two languages for training instead of only using one single language. We transform the features keyword templates and searching content to fixed-dimensional vectors calculate distances between segments obtained sliding manner. An auxiliary variability-invariant loss is also...

Domain-generalized few-shot text classification (DG-FSTC) is a new setting for (FSTC). In DG-FSTC, the model meta-trained on multi-domain dataset, and meta-tested unseen datasets with different domains. However, previous methods mostly construct semantic representations by learning from words directly, which limited in domain adaptability. this study, we enhance adaptability of utilizing distributional signatures texts that indicate domain-related features specific We propose Multi-level...

Cervical cell early detection can timely discover abnormal cells and take treatment to avoid cervical cancer. With the development of deep learning technology, smear technology has made rapid progress. However, current segmentation algorithms still have problems with inaccurate low accuracy when dealing high overlap cells. Therefore, we propose degree measure overlap, in order compare algorithm more targetedly. At same time, address misclassification problem traditional Cascade Mask RCNN...

In practical scenarios, detecting and grasping objects accurately can be very challenging due to the uncertainty of their positions orientations, as well environmental interference. Especially when target object is occluded by other objects, traditional machine vision methods have difficulty in recognizing it. To address this problem, we propose double-attention mechanism-based segmentation detection network (DAM-SGNET). DAM-SGNET a technique used for cluttered environments. It utilizes deep...

<title>Abstract</title> With the growing adoption of deep learning in AI, flaw detection bottled liquor production has become crucial to ensure product quality and consumer satisfaction. However, existing models often face issues low efficiency, particularly multi-category multi-target scenarios, struggle with integration into resource-constrained devices. To solve these challenges, this study proposes ESW-YOLO, a lightweight model optimized detect diverse flaws production. This is designed...

Generalization ability is a principal issue in the field of machine learning. Feature selection method that can improve generalization learning algorithm. Through measuring feature count measure (FCM) decision table, select which depended strongly on classification attribute. Based above, based bagging ensemble algorithm proposed. Experiment results show proposed effective to obtain rule.

In this paper, we propose to use Deep Neural Network (DNN), which has been proved be the state-of-the-art technique in speech recognition, re-estimate confidence of keyword hypotheses verification stage spoken term detection. The recognition system based on DNN outperforms that conventional Gaussian Mixture Model (GMM) but suffers from increased decoding time. When speed or indexing is critical, it seems a trade-off between performance and utilize verification. Inspired by utilization...

In order to better realize the personalized information retrieval service, key is dynamically updating user interest model. This paper puts forward a model based on ontology, First by analyzing browsing behavior in Web environment, we improve calculation method of right value. And then, according characteristics web document TF-IDF algorithm, this foundation propose and forgetting mechanism. The experiments show that can capture change interests, further accuracy satisfaction.

This paper introduces our approaches for the Mask and Breathing Sub-Challenge in Interspeech COMPARE Challenge 2020. For mask detection task, we train deep convolutional neural networks with filter-bank energies, gender-aware features, speaker-aware features. Support Vector Machines follows as back-end classifiers binary prediction on extracted embeddings. Several data augmentation schemes are used to increase quantity of training improve models' robustness, including speed perturbation,...

Proxy Re-encryption (PRE) is a useful cryptographic structure who enables semi-trusted proxy to convert ciphertext for Alice into Bob without seeing the corresponding plaintext. Although there are many PRE schemes in recent years, few of them set up based on lattice. Not only this, these lattice-based all more complicated than traditional schemes. In this paper, through study common lattice problems such as Small integer solution (SIS) and Learning with Errors (LWE), we analyze feasibility...

In this paper, an hierarchical n-gram Language model (LM) combining words and characters is explored to improve the detection of Out-of-vocabulary (OOV) in Mandarin Spoken term (STD). The LM based on a word-level LM, with character-level estimating probabilities OOV class-based way. region containing sentence be decoded detected help are derived from LM. implementation proposed approach dynamic decoder. evaluated terms Actual weighted value (ATWV) two data sets. Experiment results show that...

Meta-learning based methods prevail in few-shot text classification (FSTC). Previous perform meta-training and meta-testing on two parts of datasets the same or similar domains. Such a setting results significant limit model performance when face with data from different domains, limits generalization FSTC models. To solve above problem, this paper, we propose new setting, namely Domain-Generalized Few-Shot Text Classification (DG-FSTC). Firstly, conduct multi-domain dataset to learn...

With the rapid development of digital image processing and recognition, intelligent grid wins a great opportunity. We could be able to get information from images quickly based on computer analysis, algorithms deep learning machine learning, data extraction, model training recognition. The article firstly reviewed current status traditional grid, including broad overview specific strengths weaknesses, secondly analyzed intelligence neural network, then achieved recognition reconstruction...