Global ETD Search
Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.
Results
Showing 1 to 20 of 21 for “"Meta-Reinforcement Learning"”.
-
Efficient Meta-Reinforcement Learning
In Meta-Reinforcement Learning (meta-RL) agents are trained on a set of tasks to prepare for and learn faster in new, unseen, but related tasks. The standard practice to build training sets in meta-RL is to use dense coverage of task distributions, generating a very large set of training tasks. …
-
Fast adaptation via meta reinforcement learning
<p>Reinforcement Learning (RL) is a way to train artificial agents to autonomously interact with the world. In practice however, RL still has limitations that prohibit the deployment of RL agents in many real world settings. This is because RL takes long, typically requires human oversight, and …
-
Meta-reinforcement learning approaches to process control
Meta-learning is a branch of machine learning which trains neural network models to synthesize a wide variety of data in order to rapidly solve new problems. Many industrial processes have similar and well-understood dynamics, which suggests it is feasible to create a generalizable controller …
-
Representations in zero-shot meta-reinforcement learning
… error -- a problem well modeled by the field of Reinforcement Learning (RL). A long-standing goal in reinforcement learning is to create an agent capable of solving a diverse set of learning problems (i.e., broad generalization), and also learn quickly from few samples (i.e., rapid adaptation). …
-
Advances in meta-reinforcement learning and imitation learning
Recent advances in machine learning (ML) have led to remarkable progress in artificial intelligence (AI), enabling computers to solve previously intractable problems. While breakthroughs in generative AI, such as large language models and diffusion-based image generators, have captured public …
-
Provably Robust Offline Meta-Reinforcement Learning under Bagged Rewards via Spectral Representations
Meta-reinforcement learning trains agents to adapt to new tasks by drawing on experience from related tasks. In the offline setting, the agent must learn entirely from a fixed dataset without further interaction with the environment. A common approach, known as context-based offline …
-
Learning in the Presence of Skew and Missing Labels Through Online Ensembles and Meta-reinforcement Learning
… and social media analysis. Algorithms learning from such streams need to be able to construct near real-time models that continuously adapt to potential changes in patterns, in order to retain high performance throughout the stream. It follows that there are numerous challenges involved …
-
Adaptable and Scalable Multi-Agent Graph-Attention Communication
… of communication overhead. Within multi-agent reinforcement learning, synthesizing effective policies requires reasoning about when to communicate, whom to communicate with, and how to process messages. Meanwhile, in real-world problems, training policies and communication strategies that are …
-
Towards Scalable and Personalized Collaborative Learning
This thesis studies collaborative learning framework, where a group of agents cooperate to learn a powerful model from their local data in a distributed or decentralized manner. We address the challenges of communication overhead, data heterogeneity, scalability, and robustness by proposing and …
-
On Safety in Sequential Decision Learning
… safety challenges in sequential decision-learning systems across various applications, from recommendation systems to large language models and robotics. The research explores three key aspects of safety: per-time-step constraint satisfaction, safety in expectation, and safe adaptation to …
-
Reinforcement Learning Methods for Autonomous Spacecraft Guidance, Navigation, and Control
Machine learning is a rapidly growing field that has the potential to revolutionize many sectors of economy and research. In this context, the increased interest in machine learning for space applications is driven by its potential to enable increased autonomy and flexibility. Indeed, machine …
-
Meta-Learning Priors from Limited Data: From Theory to Practice
Unlike machine learning systems, humans can learn new concepts from only a few examples and efficiently adapt to changing circumstances. Machine learning systems typically require magnitudes more data to learn similar concepts or adapt to changes. This is because they lack domain-specific prior …
-
Meta multi-objective reinforcement learning for communication load balancing
… ones. In this thesis, we study multi-objective reinforcement learning (MORL) and meta reinforcement learning (meta-RL) for load balancing to learn highly customized policies for different trade-offs between network performance metrics. We begin with a thorough review of existing load balancing …
-
Artificial Intelligence-Driven Rate Adaptive and Perceptual-Aware Rich Media Delivery in 5G and Beyond Networks
… EDGE360, a novel edge-assisted Multi-Agent Deep Reinforcement Learning (MADRL) framework for region-aware ABR control in tile-based 360-degree video streaming. EDGE360 employs the Counterfactual Multi-Agent (COMA) policy gradient method to optimize bitrate allocation across viewport, marginal, …
-
Derivative-Free Meta-Blackbox Optimization on Manifold
… applications. This thesis presents the first meta-learning framework that leverages the shared structure among sequential tasks to improve the computational efficiency and sample complexity of derivative-free optimization. Based on the observation that most practical high-dimensional functions …
-
A Multi-Objective Deep Learning Methodology for Morphing Wings
… of a given mission profile, 2) training a Reinforcement Learning agent to modify the optimized shape and design the wing structure to ensure the structural integrity of morphing wings throughout the flight while maintaining a high aerodynamic performance 3) implementing a Meta Reinforcement …
-
Active heterogeneous graph neural networks with per-step meta-q-learning
Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2024-05-01
-
Path Planners and Controllers for Autonomous Navigation of Robots in Unstructured Domains
… solutions addressing path planning, MCAR (Meta Reinforcement-Learning based Contextual Adaptive-Control using Representations) and motion planning, RLP (Reshaping Local Path Planner). Control of non-linear dynamic systems in the presence of unknowns from a nominally narrow distribution may …
-
Enhancing Reinforcement Learning Using Data and Structure
Reinforcement learning is a powerful approach for training intelligent agents to make decisions in complex environments. However, these algorithms often struggle when faced with challenging scenarios, such as sparse reward feedback or environments with multiple learning agents. Additionally, the …
Page 1 of 2