In the evolutionary computation community, it is widely believed that stagnation impedes convergence in evolutionary algorithms, and that convergence inherently indicates optimality. However, this perspective is misle...
详细信息
Reinforcement learning (RL) has achieved significant success in task-oriented dialogue (TOD) policy learning. Nevertheless, training dialogue policy through RL faces a critical challenge: insufficient exploration, whi...
详细信息
Reinforcement learning (RL) has achieved significant success in task-oriented dialogue (TOD) policy learning. Nevertheless, training dialogue policy through RL faces a critical challenge: insufficient exploration, which leads to the policy getting trapped in local optima. evolutionary algorithms (EAs) enhance exploration breadth by maintaining and selecting diverse individuals, and they often add parameter space noise among different individuals to simulate mutation, thereby increasing exploration depth. This approach has proven to be an effective method for enhancing RL exploration and has shown promising results in game domains. However, previous research has not analyzed its effectiveness in TOD dialogue policy. Given the substantial differences between gaming contexts and TOD dialogue policy, this paper explores and validates the efficacy of EAs in TOD dialogue policy, investigating the effects of different evolutionary cycles and various noise strategies across different dialogue tasks to determine which combination of evolutionary cycle and noise strategy is most suitable for TOD dialogue policy. Additionally, we propose an adaptive noise evolution method that dynamically adjusts noise scales to improve exploration efficiency. Experiments on the MultiWOZ dataset demonstrate significant performance improvements, achieving state-of-the-art results in both on-policy and off-policy settings.
Parallel evolutionary algorithms (PEAs) have been studied for reducing the execution time of evolutionary algorithms by utilizing parallel computing. An asynchronous PEA (APEA) is a scheme of PEAs that increases compu...
详细信息
It is natural to think of evolutionary algorithms as highly stochastic search methods. This can also make evolutionary algorithms, and particularly recombination, quite difficult to analyze. One way to reduce randomne...
详细信息
ISBN:
(纸本)9781450371285
It is natural to think of evolutionary algorithms as highly stochastic search methods. This can also make evolutionary algorithms, and particularly recombination, quite difficult to analyze. One way to reduce randomness involves the quadratization of functions, which is commonly used by modern optimization methods, and also has applications in quantum computing. After a function is made quadratic, random mutation is obsolete and unnecessary;the location of improving moves can be calculated deterministically, on average in O(1) time. Seemingly impossible problems, such as the Needle-in-a-Haystack, becomes trivial to solve in quadratic form. One can also provably tunnel, or jump, between local optima and quasilocal optima in O(n) time using deterministic genetic recombination. The talk also explores how removing randomness from evolutionary algorithms might provide new insights into natural evolution. Finally, a form of evolutionary algorithm is proposed where premature convergence is impossible and the evolutionary potential of the population remains open-ended.
Scania has been working with statistics for a long time but has invested in becoming a data driven company more recently and uses data science in almost all business functions. The algorithms developed by the data sci...
详细信息
Scania has been working with statistics for a long time but has invested in becoming a data driven company more recently and uses data science in almost all business functions. The algorithms developed by the data scientists need to be optimized to be fully utilized and traditionally this is a manual and time consuming process. What this thesis investigates is if and how well evolutionary algorithms can be used to automate the optimization process. The evaluation was done by implementing and analyzing four variations of genetic algorithms with different levels of complexity and tuning parameters. The algorithm subject to optimization was XGBoost, a gradient boosted tree model, applied to data that had previously been modelled in a competition. The results show that evolutionary algorithms are applicable in finding good models but also emphasizes the importance of proper data preparation.
The Minimum Spanning Tree problem (abbr. MSTP) is a well-known combinatorial optimization problem that has been extensively studied by the researchers in the field of evolutionary computing to theoretically analyze th...
详细信息
Although differential evolution (DE) algorithms perform well on a large variety of complicated optimization problems, only a few theoretical studies are focused on the working principal of DE algorithms. To make the f...
详细信息
Many biological processes have been the source of inspiration for heuristic methods that generate high-quality solutions to solve optimization and search problems. This thesis presents an epigenetic technique for Evol...
详细信息
The design of a Compact Dual-band Equatorial helix antenna using Computational Electromagnetic Methods together with multiobjective optimization algorithms is presented. These antennas are used for Telemetry, Tracking...
详细信息
The design of a Compact Dual-band Equatorial helix antenna using Computational Electromagnetic Methods together with multiobjective optimization algorithms is presented. These antennas are used for Telemetry, Tracking, and Control of satellites from the terrain base station. In order to optimize the parameters an antenna, a simulation-optimization process is shown along a real case study. The parameters of the antenna that fulfills the radiation patterns needed for the communication are obtained using a simulation tool called MONURBS together with two well-known multiobjective algorithms: NSGA-II and SPEA-2. In this paper, a comparison with previous designs and the antenna prototype is presented, showing that this approach can obtain multiple valid solutions and accelerate the design process.
Maintaining population diversity is critical for multi-objective evolutionary algorithms (MOEAs) to solve many-objective optimization problems (MaOPs). Reference vector guided MOEAs have exhibited superiority in handl...
详细信息
Maintaining population diversity is critical for multi-objective evolutionary algorithms (MOEAs) to solve many-objective optimization problems (MaOPs). Reference vector guided MOEAs have exhibited superiority in handling this issue, where a set of well distributed reference points on a unit hyperplane are generated to construct the reference vectors. Nevertheless, the pre-defined reference vectors could not well handle MaOPs with irregular (e.g., convex, concave, degenerate, and discontinuous) Pareto fronts (PFs). In this paper, we propose two new reference vector adaptation strategies, namely Scaling of Reference Vectors (SRV) and Transformation of Solutions Location (TSL), to handle irregular PFs. Particularly, to solve an MaOP with a convex/concave PF, SRV introduces a specific center vector and adjusts the other reference vectors around it by using a scaling function. TSL transforms the location of well-diversified solutions into a set of new reference vectors to handle degenerate/discontinuous PFs. The two strategies are incorporated into three representative MOEAs based on reference vectors and tested on benchmark MaOPs. The comparison studies with other state-of-the-art algorithms demonstrate the efficiency of the new strategies. (C) 2019 Elsevier Inc. All rights reserved.
暂无评论