Article
Mathematics
Wenjuan Li, Wenying Wang, Jingsi Chen, Weidong Rao
Summary: Sufficient dimension reduction (SDR) is a useful tool for nonparametric regression with high-dimensional predictors, but many existing SDR methods rely on certain assumptions about the distribution of predictors. In this study, inspired by Wang et al., we propose a novel and effective method that combines the aggregate method and the kernel inverse regression estimation. Our proposed approach accurately estimates the dimension reduction directions and substantially improves the exhaustivity of the estimates with complex models. It is not dependent on the arrangement of slices and reduces the influence of extreme values of the response. The method performs well in both numerical examples and a real data application.
Article
Computer Science, Information Systems
Fode Zhang, Rui Li, Heng Lian
Summary: Nonparametric quantile regression is commonly used for nonlinear quantile modeling, with a kernel approach in a reproducing kernel Hilbert space framework. To address heavy computational burden with large sample sizes, a random projection approach with m << n is considered. Theoretical results show that sketched KQR achieves minimax convergence rate when m is at least as large as the effective statistical dimension of the problem.
INFORMATION SCIENCES
(2021)
Article
Automation & Control Systems
Chengbin Xuan, Feng Zhang, Hak-Keung Lam
Summary: This paper presents a method to improve the safety of agents during the exploration stage in q-learning. By introducing a safety indicator function and a safe exploration mask, the algorithm reduces the likelihood of unsafe actions and improves its applicability in industrial settings.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2022)
Article
Computer Science, Interdisciplinary Applications
Lei Wang, Puying Zhao, Jun Shao
Summary: This article introduces three different semi-parametric estimation methods to estimate distribution functions and quantiles of a response variable. An instrumental covariate is used to address the identifiability problem, and dimension reduction technique is employed to improve efficiency. The proposed estimators have been shown to be consistent and asymptotically normal.
COMPUTATIONAL STATISTICS & DATA ANALYSIS
(2021)
Article
Computer Science, Artificial Intelligence
Chao Tan, Sheng Chen, Xin Geng, Genlin Ji
Summary: In this paper, a novel label distribution manifold learning (LDML) method is proposed for accurately solving the multilabel distribution learning problem. Through manifold learning and multi-output kernel regression, accurate label distributions can be estimated and an enhanced maximum entropy model is formed. Experimental results demonstrate the advantages of the proposed LDML method in terms of learning accuracy.
PATTERN RECOGNITION
(2023)
Article
Automation & Control Systems
Samson J. Koelle, Hanyu Zhang, Marina Meila, Yu-Chia Chen
Summary: This paper studies the problem of recovering the meaning of the new low-dimensional representation in an automatic and principled fashion. A method is proposed to explain the embedding coordinates and its effectiveness is demonstrated through experiments.
JOURNAL OF MACHINE LEARNING RESEARCH
(2022)
Article
Anesthesiology
Angelos-Miltiadis Krypotos, Geert Crombez, Maryna Alves, Nathalie Claes, Johan W. S. Vlaeyen
Summary: This study investigates how individuals solve the exploration-exploitation dilemma when facing pain and finds that participants tend to choose the safest option, prioritize rewards over pain, and are more inclined to explore after experiencing pain.
Article
Computer Science, Theory & Methods
Yunxu Bai, Xinjiang Lu
Summary: In order to address the problems of excessive rule numbers and dimension disaster in the T-S fuzzy model, this study proposes a rule reduction method based on multiple kernel learning (MKLRR). By merging multiple fuzzy sets into larger sets and applying multiple kernel learning to project them from a low-dimensional space to a high-dimensional space, the final model is constructed through fuzzy inference, reducing the rule number while maintaining modeling accuracy.
FUZZY SETS AND SYSTEMS
(2023)
Article
Statistics & Probability
Eliana Christou
Summary: Marginalizing the characterizing of tail events can lead to disastrous consequences. This research explores dimension reduction techniques for conditional expectiles, introducing the central expectile subspace and a nonlinear extension approach. The algorithms are validated through extensive simulations and a real data application, demonstrating the effectiveness of expectile regression as a tool for describing tail events.
Article
Computer Science, Artificial Intelligence
Norman Tasfi, Eder Santana, Luisa Liboni, Miriam Capretz
Summary: The Successor Feature framework improves task transfer in Reinforcement Learning by decomposing the state-action value function. However, the original formulation may fail due to changes in the reward function. This paper proposes the Dynamic Successor Feature framework, DynSF, which centers around a learned state-transition model and dynamically induces the acting policy. The flexibility of DynSF extends to the architecture, requiring only a state-transition model and a small vector of parameters.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Rafael F. Katopodis, Priscila M. V. Lima, Felipe M. G. Franca
Summary: n-tuple neural networks have been widely used in various learning domains, but existing systems have limitations in terms of objective function flexibility and handling nonstationarity in online learning. A novel n-tuple system is proposed to address these issues, and its capabilities are showcased in reinforcement learning tasks.
Article
Robotics
Annie S. Chen, HyunJi Nam, Suraj Nair, Chelsea Finn
Summary: Learning from diverse offline datasets is a promising approach towards learning general purpose robotic agents. However, a core challenge lies in collecting meaningful data without human intervention. By focusing exploration on important parts of the state space with weak human supervision, the proposed Batch Exploration with Examples (BEE) technique shows improved interaction with relevant objects and task performance in vision-based robotic manipulation tasks.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Chemistry, Analytical
Ao Feng, Yuyang Xie, Yankang Sun, Xuanzhi Wang, Bin Jiang, Jian Xiao
Summary: This paper proposes a Local-and-Global Strategy (LAGS) algorithm that combines a local exploration strategy with a global perception strategy to solve the regional legacy issues in autonomous exploration, improving exploration efficiency. Experiments show that the proposed method achieves shorter paths, higher efficiencies, and stronger adaptability in exploring unknown environments.
Article
Chemistry, Analytical
Dohyun Kyoung, Yunsick Sung
Summary: This paper proposes a method to reduce the initial exploration range in reinforcement learning through a pretrained transformer decoder on expert data. The method involves pretraining a transformer decoder with massive expert data to guide the agent's actions during the early learning stages. The experiment results show that the proposed method outperforms the traditional Deep Q-Network (DQN) using the epsilon-greedy strategy in terms of average reward and win rate.
Article
Computer Science, Information Systems
Peng Yang, Qi Yang, Ke Tang, Xin Yao
Summary: Effective exploration is crucial for a successful search process. The NCNES method presented in this paper shows the importance of coordinated parallel exploration and demonstrates significant advantages, especially in games with uncertain and delayed rewards.
FRONTIERS OF COMPUTER SCIENCE
(2021)