Kalman Filter

Welch, George R.

doi:10.1007/978-3-030-63416-2_716

Cited by 35 publications

(31 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…To explicitly model the user-level preference shift and provide a better embedding initialization for graph convolution model training, we propose a temporal attention (TA) module to model the user temporal preference shift. This is motivated by the Kalman filter [3,26,57], which is an effective way to model the temporal state change. In recommender system, the user embedding e 𝑢 is utilized to represent the user 𝑢's preference.…”

Section: Temporal User Preference Modeling As Initializationmentioning

confidence: 99%

See 1 more Smart Citation

Dynamically Expandable Graph Convolution for Streaming Recommendation

Zhang

et al. 2023

Proceedings of the ACM Web Conference 2023

View full text Add to dashboard Cite

Personalized recommender systems have been widely studied and deployed to reduce information overload and satisfy users' diverse needs. However, conventional recommendation models solely conduct a one-time training-test fashion and can hardly adapt to evolving demands, considering user preference shifts and ever-increasing users and items in the real world. To tackle such challenges, the streaming recommendation is proposed and has attracted great attention recently. Among these, continual graph learning is widely regarded as a promising approach for the streaming recommendation by academia and industry. However, existing methods either rely on the historical data replay which is often not practical under increasingly strict data regulations, or can seldom solve the overstability issue. To overcome these difficulties, we propose a novel Dynamically Expandable Graph Convolution (DEGC) algorithm from a model isolation perspective for the streaming recommendation which is orthogonal to previous methods. Based on the motivation of disentangling outdated short-term preferences from useful long-term preferences, we design a sequence of operations including graph convolution pruning, refining, and expanding to only preserve beneficial long-term preference-related parameters and extract fresh short-term preferences. Moreover, we model the temporal user preference, which is utilized as user embedding initialization, for better capturing the individual-level preference shifts. Extensive experiments on the three most representative GCN-based recommendation models and four industrial datasets demonstrate the effectiveness and robustness of our method.

show abstract

Section: Temporal User Preference Modeling As Initializationmentioning

confidence: 99%

“…The added filters will also be partly pruned to eliminate the redundant ones and prevent the network width explosion catastrophe. Moreover, inspired by the Kalman filter [57], a temporal attention model is utilized to explicitly encode the temporal user preference, which works as the user embedding initialization for training on new data.…”

Section: Introductionmentioning

confidence: 99%

Dynamically Expandable Graph Convolution for Streaming Recommendation

Zhang

et al. 2023

Proceedings of the ACM Web Conference 2023

View full text Add to dashboard Cite

show abstract

“…After the radar or photoelectric system finds a target, it converts the target information into information in the geodetic coordinate system, and uses the Kalman filter algorithm [13] to predict the position. Spectral clustering DBSCAN algorithm [12] is used to cluster the target, and the clustering result is a bee colony.…”

Section: Shooting Strategy Frameworkmentioning

confidence: 99%

Development of a shooting strategy to neutralize UAV swarms based on multi-shot cooperation

Shaohui,

Ji,

Baohua

et al. 2023

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

Economical and efficient assault methods using drones and drone swarms severely assailed the defenders during the conflict in the Middle East. In view of the current situation, based on medium and large-caliber anti-aircraft artillery platforms, this article analyzes and calculates the advantages of using multiple anti-aircraft guns against drones, the threat of a drone colony, and the drone colony density. Additionally, this study determines the firing sequence of a drone swarm, and provides anti-aircraft gun firing position points to reduce the firing cost and increase the efficiency. This ensures the optimization of an air defense system within a short duration of time. Ammunition is an effective defense against drone swarms. The experimental results demonstrate that compared with that of the conventional methods of individually shooting the swarms, this strategy can simultaneously neutralize multiple targets, assign shooting tasks to multiple artillery guns, and can cause greater damage to a drone swarm.

show abstract

“…For a more detailed discussion on Kalman filters we refer the interested reader to the literature [133] . Other (nonlinear) ways to perform inference in SSMs exist, for instance using sequential Monte Carlo estimation [134] .…”

Section: State-space Modelsmentioning

confidence: 99%

Data-driven science and machine learning methods in laser–plasma physics

Döpp

Eberle

Howard

et al. 2023

High Pow Laser Sci Eng

View full text Add to dashboard Cite

Laser-plasma physics has developed rapidly over the past few decades as lasers have become both more powerful and more widely available. Early experimental and numerical research in this field was dominated by single-shot experiments with limited parameter exploration. However, recent technological improvements make it possible to gather data for hundreds or thousands of different settings in both experiments and simulations. This has sparked interest in using advanced techniques from mathematics, statistics and computer science to deal with, and benefit from, big data. At the same time, sophisticated modeling techniques also provide new ways for researchers to deal effectively with situation where still only sparse data are available. This paper aims to present an overview of relevant machine learning methods with focus on applicability to laser-plasma physics and its important sub-fields of laser-plasma acceleration and inertial confinement fusion.

show abstract

Kalman Filter

Cited by 35 publications

References 6 publications

Dynamically Expandable Graph Convolution for Streaming Recommendation

Dynamically Expandable Graph Convolution for Streaming Recommendation

Development of a shooting strategy to neutralize UAV swarms based on multi-shot cooperation

Data-driven science and machine learning methods in laser–plasma physics

Contact Info

Product

Resources

About