Traffic speed prediction based on real-world traffic data is a classical problem in intelligent transportation systems (ITS). Most existing models are proposed the hypothesis that complete or have rare missing values. However, such collected scenarios often incomplete due to various human and natural factors. Although this can be solved by first estimating values with an imputation model then applying model, former potentially breaks critical latent features further leads error accumulation...

Medical time series has been playing a vital role in real-world healthcare systems as valuable information monitoring health conditions of patients. Accurate classification for medical series, e.g., Electrocardiography (ECG) signals, can help early detection and diagnosis. Traditional methods towards rely on handcrafted feature extraction statistical methods; with the recent advancement artificial intelligence, machine learning deep have become more popular. However, existing often fail to...

Diffusion-based generative models have recently excelled in generating molecular conformations but struggled with the generalization issue -- trained on one dataset may produce meaningless out-of-distribution molecules. On other hand, distance geometry serves as a generalizable tool for traditional computational chemistry methods of conformation, which is predicated assumption that it possible to adequately define set all potential any non-rigid system using purely geometric constraints. In...

Statistics on urban traffic speed flows are essential for thoughtful city planning. Recently, data-driven prediction methods have become the state-of-the-art a wide range of forecasting tasks. However, many small cities limited amount data available building models due to lack collection methods. With acceleration urbanization, need construction and medium-sized is imminent. To tackle above problems, we propose TransfEr lEarning approach with graPh nEural nEtworks (TEEPEE) that can forecast...

Digital humanities is an important subject because it enables developments in history, literature, and films. In this article, we perform empirical study of a Chinese historical text, Records the Three Kingdoms (Records), novel same story, Romance (Romance). We employ deep-learning-based natural language processing (NLP) techniques to extract characters their relationships. The adopted NLP approach can 93% 91% that appeared two books, respectively. Then, characterize social networks...

How to obtain accurate travel time predictions is among the most critical problems in Intelligent Transportation Systems (ITS). Recent literature has shown effectiveness of machine learning models on forecasting problems. However, these predict a point estimation manner, which not suitable for real scenarios. Instead determined value, within future period distribution. Besides, they all use grid structure data spatial dependency, does reflect traffic network's actual topology. Hence, we...

In modern traffic management, one of the most essential yet challenging tasks is accurately and timely predicting traffic. It has been well investigated examined that deep learning-based Spatio-temporal models have an edge when exploiting relationships in data. Typically, data-driven require vast volumes data, but gathering data small cities can be difficult owing to constraints such as equipment deployment maintenance costs. To resolve this problem, we propose TrafficTL, a cross-city...

Understanding how people choose to travel is essential for intelligent transportation planning and related smart services. Recent advances in deep learning, coupled with the increasing market penetration of GPS devices, have paved way novel mode identification methods based on data mining. While many shown promising results, most often relied heavily few available labeled data, leaving large amounts unlabeled ones unused. To address this issue, we propose MultiMix, a semi-supervised...

We introduce a pioneering methodology for boosting large language models in the domain of protein representation learning. Our primary contribution lies refinement process correlating over-reliance on co-evolution knowledge, way that networks are trained to distill invaluable insights from negative samples, constituted by pairs sourced disparate categories. By capitalizing this novel approach, our technique steers training transformer-based within attention score space. This advanced...

Tabular data have been playing a mostly important role in diverse real-world fields, such as healthcare, engineering, finance, etc. With the recent success of deep learning, many tabular machine learning (ML) methods based on networks (e.g., Transformer, ResNet) achieved competitive performance benchmarks. However, existing ML suffer from representation entanglement and localization, which largely hinders their prediction leads to inconsistency tasks. To overcome these problems, we explore...

In modern traffic management, one of the most essential yet challenging tasks is accurately and timely predicting traffic. It has been well investigated examined that deep learning-based Spatio-temporal models have an edge when exploiting relationships in data. Typically, data-driven require vast volumes data, but gathering data small cities can be difficult owing to constraints such as equipment deployment maintenance costs. To resolve this problem, we propose TrafficTL, a cross-city...

Travel time estimation (TTE) aims to predict travel duration and provide reliable planning for residential schedules. Trajectories naturally contain sequential features in form of GPS points with temporal precedence, which can be leveraged improve prediction performance. Besides, the spatial information, i.e. graph structure road network, well represent highly is commonly used capture information traffic networks. However, extracting regional from trajectory data, addition its latitude...