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 126 for “"Inductive Biases"”.
-
Inductive Biases in Video Understanding
Inductive biases are the assumptions made in a machine learning algorithm, typically introduced to restrict the hypothesis space and improve learning. For computer vision models, these biases are found in the functions for image or video processing, methods for producing outputs, and the …
-
ALGORITHMIC INDUCTIVE BIASES FOR GRAPH REPRESENTATION LEARNING
… in the neural network that can provide useful inductive bias in learning. However, introducing useful structures in the neural network can often be non-trivial and may need significant domain knowledge. In this thesis, we propose to use the knowledge of well-known algorithms to provide the …
-
Incorporating inductive biases into machine learning algorithms
… address this challenge from the perspective of inductive biases. Specifically, we investigate how to effectively use human knowledge about data or tasks to optimize the behavior of a machine learning algorithm, without requiring extra data. We will first give a brief review of research on …
-
Inductive Biases for Higher-Order Visual Cognition
… based machine learning models is the use of inductive biases. Inductive biases allow human designers to incorporate knowledge about the learning task into the models, the data used to train them, as well as their training regime. These biases enable DNNs to learn structured representations of …
-
On the inductive biases of deep generative models
… available data at inference time. However, the inductive biases why generative models work well, and their critical components such as their neural architecture are often poorly understood.<br><br>In this thesis, we develop, characterise and probe the inductive biases of four major generative …
-
Causal inductive biases for learning Structured World Models
World Models have emerged as a powerful paradigm for building autonomous agents: by learning a predictive model of the environment, agents can plan and make decisions based on imagined future scenarios. Despite their success, existing approaches to learning world models largely rely on dense, …
-
Inductive biases and generalisation for deep reinforcement learning
… incorporating additional structures, also called inductive biases, into the agent. Focussing on specific, yet widely applicable problem domains, we can develop specialised architectures which greatly improve performance. In Chapter 3 we focus on partially observable environments in which the agent …
-
Scalable syntactic inductive biases for neural language models
… for augmenting neural sequence models with an inductive bias that encourages a more explicit modelling of hierarchical syntactic structures. In the first approach, we use existing techniques that design <strong>tree-structured</strong> neural networks, where the ordering of the computational …
-
Inductive biases and metaknowledge representations for search-based optimization
… problem-solving in terms of domain-specific inductive biases and the arbitration thereof through means of high-level abstractions resolves outstanding issues with methodology proposed within the literature. Noteworthy, it can be also related to ongoing research on algorithm selection and …
-
On Using Inductive Biases for Designing Deep Learning Architectures
… takes our prior knowledge of the system as an inductive bias? In this dissertation, we develop two novel approaches towards this end. The first one called `Cooperative Neural Networks' can incorporate the inductive bias from the underlying probabilistic graphical model representation of the …
-
Implicit Regularization and Inductive Biases in Neural Network Training
… this challenge: implicit regularization and inductive bias. Implicit regularization refers to the tendency of the training process itself --- through optimization algorithms, parameterization, and initialization --- to favor certain solutions over others, even in the absence of explicit …
-
Inductive Biases in Field-Level Cosmological Inference from Galaxy Catalogs
… using machine learning models with differing inductive biases. Using features extracted from hydrodynamic simulations in the CAMELS suite, we investigate how both observable choice and model architecture govern the extraction of cosmological information. We consider galaxy positions and …
-
Stronger Inductive Biases for Sample-Efficient and Controllable Neural Machine Translation
… NMT models by incorporating stronger inductive biases that help correct undesirable biases, integrate prior knowledge, and introduce flexible ways to control the outputs in NMT. In our first line of research, we show that current NMT models are susceptible to undesirable biases that …
-
BRIDGING INTERPRETABILITY AND PERFORMANCE IN 3D DEEP LEARNING THROUGH GEOMETRIC INDUCTIVE BIASES
Questa tesi indaga come i geometric inductive biases possano risolvere limitazioni fondamentali nel deep learning 3D, in particolare per applicazioni critiche di sicurezza come l’ispezione delle reti elettriche. Nonostante i progressi significativi nel campo del 3D scene understanding, la maggior …
-
Equivariant Inductive Biases for Weather Prediction with PEAR - Investigating the exploitation of rotational symmetries for accurate transformer-based weather forecasting over the HEALPix discretisation
… outcome highlights the difficulty of embedding inductive biases in the case of domains that involve using high-dimensional samples, specifically in relation to window-based attention mechanisms.
-
Domain Knowledge as Motion-Aware Inductive Bias for Deep Video Synthesis: Two Case Studies
… arguably due to the lack of sufficiently strong inductive biases that encourage the models to capture the temporal-dynamic nature of video data. Inductive bias refers to the prior knowledge incorporated into the learning models to explicitly drive the learning process toward the solutions that …
-
ENHANCING NEURAL NETWORKS FOR TSP AND VRP: ALGORITHMIC BIAS, HIERARCHICAL STRUCTURES, AND FOUNDATION MODELS FOR EFFICIENT ROUTING
… This thesis investigates how embedding effective inductive biases into neural networks can improve their ability to solve the TSP and VRP, particularly under realistic problem distributions. We present three complementary approaches. First, we study algorithmic inductive bias by integrating …
-
Disentangling neural network representations for improved generalization
… in data collection and model design. Generally, inductive biases can make this process easier by leveraging knowledge about the world to guide neural network design. One such inductive bias is disentanglment, which can help preven neural networks from learning representations that capture …
-
Generalizing speech recognition techniques for monotonic sequence modeling applications
… in a larger system. These features for encoding inductive biases about our problem improve data efficiency, which is especially beneficial when paired data for learning the end-to-end problem from scratch is insufficiently plentiful or entirely unavailable. We can thus get the best of both …
-
New Algorithms for Attribute-Efficient on -Line Linear Learning
… goal of this work was to identify the inductive biases of each algorithm, so that they can be fairlycompared with each other. By examining their biases and properties using the results presented here, it is possible to view 2Pes as a particular generalization of the Winnow algorithm, …
Page 1 of 7