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 13 of 13 for “"neural process"”.
-
Advances in Probabilistic Meta-Learning and the Neural Process Family
… remains a significant bottleneck in deploying neural network based solutions in several important application domains. But how can we reason about and design solutions to this daunting task? This thesis is concerned with a particular perspective for meta-learning in supervised settings. We view …
-
Convolutional Conditional Neural Processes
Neural processes are a family of models which use neural networks to directly parametrise a map from data sets to predictions. Directly parametrising this map enables the use of expressive neural networks in small-data problems where neural networks would traditionally overfit. Neural processes can …
-
The Neural Processes Family: Translation Equivariance and Output Dependencies
… 2021; Schmidhuber, 1987) directly from data. Neural processes (Garnelo et al., 2018a,b) are a family of meta-learning models which combine the flexibility of deep learning with the uncertainty awareness of probabilistic models. Training using meta-learning allows neural processes to apply deep …
-
Approximate Inference in Bayesian Neural Networks and Translation Equivariant Neural Processes
… probabilistic machine learning models: Bayesian neural networks and neural processes. Bayesian neural networks are a classical model that has been the subject of research since the 1990s. They rely on Bayesian inference to represent uncertainty in the weights of a neural network. On the other …
-
Working memory, selective visual attention and hierarchical perception
… reflects on whether or not this top-down process of attentional capture from working memory is an automatic mechanism where attention gets deployed without a need for voluntary effort, and on the neural process of this endogenous control working in conjunction with bottom-up exogenous …
-
Active Inference in Multi-Objective Dynamic Environments
… viability in modelling scenarios both related to neural process theory and more classical agent-based machine learning. However, due to the relative recency of the theory, there are still many areas of comparison and evaluation to explore. This dissertation aims to investigate Active Inference's …
-
Bottom-up and Top-down Mechanisms of Visually-Guided Movements
<p>Interacting with the world is a two-step process of accurate sensing followed by coordinated movement. Optimization of biologically-inspired robotic systems benefits from the quantification and modeling of natural sensorimotor behavior, including the bottom-up circuits that mediate it and …
-
Attentional Mechanisms in Natural Scenes
… of the world around us is an incredibly complex neural process that allows humans to function appropriately within the environment. When one considers the intricacy of both the visual input and the (currently known) neural mechanisms necessary for its analysis, it is difficult not to remain …
-
Exploiting multimodality and structure in world representations
… scaled to much larger graphs than the ones processed by the best-performing method at the time, or incorporated theoretical properties via the use of topological data analysis algorithms. Both approaches competed with contemporary state-of-the-art graph classification methods, even outside …
-
Machine Learning Approaches to Assessing Future Flood & Storm Risk
… or hidden states, we modify and extend the Neural Process framework, examining the structure of the encoder and decoder networks within. Whilst we achieve state of the art performance with the first foray, using Recurrent Neural Networks as encoder and decoder, our second, using Temporal …
-
Exploiting the Intrinsic Embodied Dynamics for Adaptive Robotic Behaviors
… emerged with a reduced computation load on the neural process. This technique enables compliant interactions with the environment, allowing for more adaptive behaviors in complex and unstructured environments. This thesis explores how the intrinsic embodied dynamics can be utilized for the …
-
Neural Dynamics of Goal-Directed Action Selection: A Multivariate Approach to Evidence Accumulation and Cognitive Control in Response to Competing Stimuli
… for goal-directed behaviour might arise from neural activity—the capacity to integrate information about a stimulus to inform action-selection, and the capacity to resolve conflict in the presence of competing alternatives. I examine this over the course of three empirical studies …
-
Data and Computation Efficient Meta-Learning
… this, we introduce CNAPs, a conditional neural process based approach to multi-task classification. We demonstrate that, at the time, CNAPs achieved state-of-the-art results on the challenging Meta-Dataset benchmark indicating high-quality transfer-learning. Timing experiments reveal that …