We advance the study of incentivized bandit exploration, in which arm choices are viewed as recommendations and are required to be Bayesian incentive compatible. Recent work of (Sellke & Slivkins, 2022) has shown ...
详细信息
We advance the study of incentivized bandit exploration, in which arm choices are viewed as recommendations and are required to be Bayesian incentive compatible. Recent work of (Sellke & Slivkins, 2022) has shown that for the special case of independent arms, after collecting enough initial samples, the popular Thompson sampling algorithm becomes incentive compatible. This was generalized to the combinatorial semibandit in (Hu et al., 2022). We give an analog of this result for linear bandits, where the independence of the prior is replaced by a natural convexity condition. This opens up the possibility of efficient and regret-optimal incentivized exploration in high-dimensional action spaces. In the semibandit model, we also improve the sample complexity for the pre-Thompson sampling phase of initial data collection.
The expansion of urbanization leads to significant changes in land use, consequently affecting carbon storage. This research aims to investigate the carbon loss due to land use alterations and proposes strategies for ...
详细信息
ISBN:
(纸本)9789887891826
The expansion of urbanization leads to significant changes in land use, consequently affecting carbon storage. This research aims to investigate the carbon loss due to land use alterations and proposes strategies for mitigation. Utilizing existing land use data from 2017 and 2022, along with simulated data for 2025 generated by an ANN model and Cellular Automata, we identified changes in land use. These changes were then correlated with variations in carbon storage, both gains and losses. Our findings reveal a significant loss of 36,859 metric tons of carbon storage from 2017 to 2022. The projection for 2025 estimates a further reduction, reaching a total loss of 83,409 metric tons. By employing the LISA method, we identified that low-carbon storage zones are concentrated in the southeast region of the research site. By overlaying these zones with areas of carbon storage loss, we pinpointed regions severely affected by carbon depletion. Consequently, we propose that mitigation strategies should be imperatively implemented in these identified areas to counteract the trend of carbon storage loss. This approach offers urban planners a solution to identify areas experiencing carbon storage decline. Moreover, our research methodology provides a novel framework for scholars studying similar carbon issues.
In this research, we present a novel solution to the problem of multi-robot task allocation (MRTA) by utilizing a variant of the state-of-art Deep Reinforcement learning algorithm, Soft Actor-Critic (SAC) within Multi...
详细信息
Skin cancer is one of the most prevalent forms of cancer around the world. Initial diagnosis relies on visual assessment of the affected area, followed by detailed dermoscopic analysis. The development of an automated...
详细信息
This paper presents the development process of a natural language to SQL model using the T5 model as the basis. The models, developed in August 2022 for an online transaction processing system and a data warehouse, ha...
详细信息
ISBN:
(纸本)9798350358810;9798350358803
This paper presents the development process of a natural language to SQL model using the T5 model as the basis. The models, developed in August 2022 for an online transaction processing system and a data warehouse, have a 73% and 84% exact match accuracy respectively. These models, in conjunction with other work completed in the research project, were implemented for several companies and used successfully on a daily basis. The approach used in the model development could be implemented in a similar fashion for other database environments and with a more powerful pre-trained language model.
In deep -sea mining, the accurate and rapid prediction of the pressure drop in a solid-liquid two-phase pipe flow (SLPF) with different parameters including particles, pipes, and flow fields, remains an issue yet to b...
详细信息
In deep -sea mining, the accurate and rapid prediction of the pressure drop in a solid-liquid two-phase pipe flow (SLPF) with different parameters including particles, pipes, and flow fields, remains an issue yet to be fully resolved. In this study, an extensive investigation of the pressure drop in a slpf is conducted using machine-learning techniques. By collecting 1290 sets of data from 13 experimental papers and performing analysis and processing, we obtain a machine-learning ensemble algorithm capable of accurately predicting the pipe -pressure drop based on random forest (RF), back propagation (BP), and polynomial regression (PR) algorithms. The performance of the ensemble algorithm surpasses that of the other three algorithms, whether applied to pure substance (PS) particles or mixed particles (MP) containing PS and equivalent particles. For PS particles, the particle concentration and particle diameter -to -pipe diameter (PTP) account for the second and third weights influencing the pressure drop. Using the computational fluid dynamics (CFD)-discrete element method (DEM), this can be attributed to the significant kinetic energy loss caused by the collisions and friction between the particles and pipe wall and the excessive gravity of the particles, which influences the pressure drop.
Groundwater is best portrayed as the world's truly covered up treasure. It has had an effect in giving safe drinking water during dry season. It is the world's biggest open store of fresh water (barring ice sh...
详细信息
Federated learning (FL) is a collaborative learning paradigm where multiple clients are used to build the model without sharing data and preserving privacy. An FL-based linear regression model is designed to predict t...
详细信息
In the expanding realms of the Internet and IoT, the surge in network data both drives the digital economy and intensifies cybersecurity vulnerabilities. Network anomaly detection is essential for protecting against s...
详细信息
ISBN:
(纸本)9798350354638;9798350354621
In the expanding realms of the Internet and IoT, the surge in network data both drives the digital economy and intensifies cybersecurity vulnerabilities. Network anomaly detection is essential for protecting against security threats. This paper conducted a comprehensive comparative study by applying big data analytics and sophisticated machinelearning to enhance intelligent network anomaly detection. It confronts challenges such as data heterogeneity and model standardization, conducting extensive experiments across six datasets with a range of algorithms, from classical decision trees to cutting-edge CNN, LSTM, and Transformer models, the GWO algorithm was also employed for feature selection, and it was combined with the KNN algorithm to optimize classification performance. The evaluation focuses on metrics such as accuracy, recall rate, F1 score, and training time, revealing the performance of these algorithms with high-dimensional and imbalanced data. Notably, random forest shows exceptional detection performance on the CIC-MalMem-2022dataset, while random forest and GBDT excel in accuracy and training speed on the RT-IoT2022dataset. These insights are critical for creating more effective detection systems.
In order to effectively reduce the risk of fire occurrence, research on fire risk prediction has been rapidly developing. However, current fire risk prediction mainly relies on the experience of regulatory personnel, ...
详细信息
ISBN:
(纸本)9798400710353
In order to effectively reduce the risk of fire occurrence, research on fire risk prediction has been rapidly developing. However, current fire risk prediction mainly relies on the experience of regulatory personnel, lacking scientific and systematic methods. To address this issue, this study utilizes the random forest model to accomplish fire prediction by analyzing multi-source fire data with smoke and flame information. Experimental results indicate that the prediction performance of the random forest model is more superior on smoke data to that on flame data, suggesting that smoke information might be a more promising indicator for fire alarming.
暂无评论