Modeling trajectory data is a building block for many smart-mobility initiatives. Existing approaches apply shallow models such as Markov chain and inverse reinforcement learning to model trajectories, which cannot capture the long-term dependencies. On other hand, deep Recurrent Neural Network (RNN) have demonstrated their strength of modeling variable length sequences. However, directly adopting RNN trajectories not appropriate because unique topological constraints faced by trajectories....

ABSTRACT Supply chain finance (SCF) platforms have become increasingly important in providing financial services within supply chains. However, their role creating competitive advantage is not well understood. This paper presents findings from an in‐depth multiple‐case study of four representative SCF service‐providing China, aiming to examine why firms build platforms, the innovative strategies adopt develop solutions, and impact these solutions on business models advantage. Data were...

Similarity search on time series is an essential operation in many applications. In the state-of-the-art methods, such as R-tree based SAX and iSAX, are by default divided into equi-length segments globally, that is, all segmented same way. Those methods then focus how to approximate or symbolize construct indexes. this paper, we make important observation: global segmentation of may incur unnecessary cost space for indexing series. We develop DSTree, a data adaptive dynamic index addition...

In order to understand a complex system, we analyze its output or log data. For example, track system's resource consumption (CPU, memory, message queues of different types, etc) help avert system failures; examine economic indicators assess the severity recession; monitor patient's heart rate EEG for disease diagnosis. Time series data is involved in many such applications. Much work has been devoted pattern discovery from time data, but not much attempted use unveil internal dynamics. this...

Vehicle trajectories are one of the most important data in location-based services. The quality directly affects However, real applications, trajectory not always sampled densely. In this paper, we study problem recovering entire route between two distant consecutive locations a trajectory. Most existing works solve without using those informative historical or it an empirical way. We claim that data-driven and probabilistic approach is actually more suitable as long sparsity can be well...

Recently, time series classification with shapelets, due to their high discriminative ability and good interpretability, has attracted considerable interests within the research community. Previously, shapelet generating approaches extracted shapelets from training or learned many parameters. Although they can achieve higher accuracy than other approaches, still confront some challenges. First, searching learning in raw space incurs a huge computation cost. For example, it may cost several...

Recently, much study has been directed toward summarizing event data, in the hope that summary will lead us to a better understanding of system generates events. However, instead offering global picture system, obtained by most current approaches are piecewise, each describing an isolated snapshot system. We argue best summary, both terms its minimal description length and interpretability, is one with internal dynamics Such includes, for example, what states how alternates among these...

Data series indexes are necessary for managing and analyzing the increasing amounts of data collections that nowadays available. These support both exact approximate similarity search, with search providing high-quality results within milliseconds, which makes it very attractive certain modern applications. Reducing pre-processing (i.e., index building) time improving accuracy two major challenges. DSTree iSAX family state-of-the-art solutions this problem. However, suffers from long...

Model selection is a critical step in time series forecasting, traditionally requiring extensive performance evaluations across various datasets. Meta-learning approaches aim to automate this process, but they typically depend on pre-constructed matrices, which are costly build. In work, we propose leverage Large Language Models (LLMs) as lightweight alternative for model selection. Our method eliminates the need explicit matrices by utilizing inherent knowledge and reasoning capabilities of...

Voxels are among the most popular 3D geometric representations today. Due to their intuitiveness and ease-of-editing, voxels have been widely adopted in stylized games low-cost independent games. However, high storage cost of voxels, along with significant time overhead associated large-scale voxel rendering, limits further development open-world In this paper, we introduce Aokana , a GPU-Driven V o xel Rendering Framewor k for Ope n World G mes. is based on Sparse Voxel Directed Acyclic...

The volume of time series data has exploded due to the popularity new applications, such as center management and IoT. Subsequence matching is a fundamental task in mining data. All index-based approaches only consider raw subsequence (RSM) do not support normalization. UCR Suite can deal with normalized match problem (NSM), but it needs scan full series. In this paper, we propose novel problem, named constrained (cNSM), which adds some constraints NSM problem. cNSM provides knob flexibly...

Fat tissue is viewed as an active endocrine organ that secretes a variety of bioactive substances. Resistin, adipocyte-secreted factor, thought to be closely related obesity, insulin resistance and inflammation, the three most significant risk factors for progression pancreatic cancer. However, association between resistin cancer still unknown. In this study, tumor samples from 45 patients with ductal adenocarcinoma were analyzed immunohistochemistry expression resistin. The correlation...

System logs are vital for diagnosing system failures, with log parsing converting unstructured into structured data. Existing methods fall two categories: non-deep-learning approaches cluster based on stats but often miss semantic information, resulting in poor performance. Deep-learning excel at identifying variables and constants lack generalizability beyond training And they always suffer from low efficiency. This paper proposes a novel LLM-based approach, named Hooglle, to address these...

Similarity matching is one of the most important operations for data mining over time series. But previous works mainly focus on certain data. With development internet things and sensor networks, uncertain series are emerging from various sources, which a new challenge processing. In this paper, novel similarity algorithm proposed based simple model representing According to certainty query database, classified three types. Then extracted represent original Finally, search adopted....

Many IoT (Internet of Things) applications, like the industrial internet and smart city, collect data continuously from massive sensors. It is crucial to exploit analyze time series efficiently. Subsequence matching a fundamental task in mining data. Most existing works develop index approach for static However, applications need continuous new deposit huge historical data, which pose significant challenge indexing approach. To address this challenge, we propose lightweight structure,...