Milvus

Wang, Jianguo; Yi, Xiaomeng; Guo, Rentong; Jin, Hai; Xu, Peng; Li, Shengjun; Wang, Xiangyu; Guo, Xiangzhou; Li, Chengming; Xu, Xiaoyu; Yu, Kun; Yuan, Yuxing; Zou, Yinghao; Long, Jiquan; Cai, Yi; Li, Zhenxiang; Zhang, Zhifeng; Mo, Yihua; Jun, Gu; Jiang, Ruiyi; Wei, Yi; Xie, Charles

doi:10.1145/3448016.3457550

Cited by 72 publications

(29 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition to feeding the input data to the model, the embedding-based method has to search for the nearest neighbors in the user movements database. Even though efficient data structures for nearest neighbor search exist [28], there is a space and time overhead to store and find the most similar items, compared to the classification-based model that directly predicts the user from the input. On the other hand, classification-based models need to be retrained when enrolling users or when the enrollment data changes in general, which is a much larger overhead.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Comparison of Data Encodings and Machine Learning Architectures for User Identification on Arbitrary Motion Sequences

Schell

Hotho

Latoschik

2022

2022 IEEE International Conference on Artificial Intelligence and Virtual Reality (AIVR)

View full text Add to dashboard Cite

Recently emerged solutions demonstrate that the movements of users interacting with extended reality (XR) applications carry identifying information and can be leveraged for identification. While such solutions can identify XR users within a few seconds, current systems require one or the other trade-off: either they apply simple distance-based approaches that can only be used for specific predetermined motions. Or they use classification-based approaches that use more powerful machine learning models and thus also work for arbitrary motions, but require full retraining to enroll new users, which can be prohibitively expensive. In this paper, we propose to combine the strengths of both approaches by using an embedding-based approach that leverages deep metric learning. We train the model on a dataset of users playing the VR game "Half-Life: Alyx" and conduct multiple experiments and analyses. The results show that the embedding-based method 1) is able to identify new users from non-specific movements using only a few minutes of reference data, 2) can enroll new users within seconds, while retraining a comparable classification-based approach takes almost a day, 3) is more reliable than a baseline classification-based approach when only little reference data is available, 4) can be used to identify new users from another dataset recorded with different VR devices. Altogether, our solution is a foundation for easily extensible XR user identification systems, applicable even to non-specific movements. It also paves the way for production-ready models that could be used by XR practitioners without the requirements of expertise, hardware, or data for training deep learning models.

show abstract

Section: Discussionmentioning

confidence: 99%

“…These are stored in a suitable data structure that allows for fast nearestneighbor search using the Euclidean distance. Examples for data structures are vector similarity databases such as Milvus [28]. We stress that while this data needs to be collected to later identify these users, the data is not used to retrain the model.…”

Section: Predictionmentioning

confidence: 99%

Comparison of Data Encodings and Machine Learning Architectures for User Identification on Arbitrary Motion Sequences

Schell

Hotho

Latoschik

2022

2022 IEEE International Conference on Artificial Intelligence and Virtual Reality (AIVR)

View full text Add to dashboard Cite

show abstract

“…Then create the interfaces for video upload, video classification, and video recommendation. Create a Milvus [17] database, second, to hold information about video features. Use Springboot to build the Web backend manager, which will implement the upload, categorization, and recommendation processes for videos.…”

Section: General System Design Overviewmentioning

confidence: 99%

A winter sports video classification method and system based on 3D-video Swin transformer

Yuan

2023

International Conference on Optics and Machine Vision (ICOMV 2023)

View full text Add to dashboard Cite

Videos of winter sports can be a very good resource for learning about sports in China. In this paper, we address the issues of a small winter sports video dataset and low classification accuracy by proposing a fast video classification approach that combines a transformer and a convolutional neural network. The 3D-Video Swin Transformer model is built using the resnet3D feature extraction component and the Video Swin Transformer, and it improves local and global modeling capabilities through a multi-headed self-attention mechanism. Convolutional operations are used at the network front-end to make for the Transformer's lack of inductive bias, enhancing the network's local modeling capabilities. Convolution operations are utilized at the network's front end to make for Transformer's lack of inductive bias, hence increasing the network's capacity for local modeling and decreasing the model's dependency on vast amounts of data. The experimental results indicate that the 3D-Video Swin Transformer model may achieve an accuracy of up to 76.43 percent on the winter sports video dataset developed in this paper. The classification impact is also substantially stronger, and this accuracy is also 1.15 percent higher than that of the Video Swin Transformer Network. Additionally, we develop and implement a winter sports video classification system based on the Milvus database to facilitate user interaction and enable the submission, categorization, and recommendation of winter sports movies.

show abstract

“…It is increasingly common that rich, unstructured data such as large texts, images and video are not only stored, but given semantics through a process called vectorization [1] which captures the features of the data object in a cost-effectively processed numerical vector such as ⃗ k = [6,7]. The vectors are n-dimensional, and consist of natural, real, or complex numbers, where one number represents a feature or a part of a feature.…”

Section: Introductionmentioning

confidence: 99%

“…The DBMS, however, in all cases, facilitates data management that is feasible, freeing development resources towards other business domain critical tasks by providing ready-made features such as transaction and access control, automated database scalability, and query optimization. Additionally, increasingly complex business domains require increasingly complex features such as vector similarity search complemented by metadata filters, as well as searching with multiple query vectors [1], and efficient ways to manage access control and concurrent transactions.…”

Section: Introductionmentioning

confidence: 99%

Error messages in relational database management systems: A comparison of effectiveness, usefulness, and user confidence

Taipalus

Grahn

Ghanbari

2021

Journal of Systems and Software

View full text Add to dashboard Cite

Milvus

Cited by 72 publications

References 34 publications

Comparison of Data Encodings and Machine Learning Architectures for User Identification on Arbitrary Motion Sequences

Comparison of Data Encodings and Machine Learning Architectures for User Identification on Arbitrary Motion Sequences

A winter sports video classification method and system based on 3D-video Swin transformer

Error messages in relational database management systems: A comparison of effectiveness, usefulness, and user confidence

Contact Info

Product

Resources

About