Laurence Aitchison
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Laurence Aitchison has a total of 73 co-authors. Here are the top co-authors by number of shared publications:
Adrià Garriga-Alonso
4 shared publicationsSebastian W. Ober
4 shared publications- Ben Anson
School of Mathematics School of Mathematics Department of Computer Science University of Bristol University of Bristol University of Bristol
3 shared publications - Xi Wang
University of Sheffield, Sheffield, United Kingdom
3 shared publications - Ali Unlu
Department of Infomatics University of Sussex Brighton UK
2 shared publications - Edward Milsom
School of Mathematics School of Mathematics Department of Computer Science University of Bristol University of Bristol University of Bristol
2 shared publications
Latest Publications
- 1 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Proceedings of the IEEE
- Vol. 114 (3) , 484-515
Human-Inspired Perspectives: A Survey on AI Long-Term Memory
- 1 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Proceedings of the IEEE
- Vol. 114 (3) , 484-515
Human-Inspired Perspectives: A Survey on AI Long-Term Memory
- Open Access
Controlling changes to attention logits
Stability of neural network weights is critical when training transformer models. The query and key weights are particularly problematic, as they tend to grow large without any intervention. Applying normalization to queries and keys, known as `QK norm', fixes stability issues in practice, but is not always applicable. For example, QK norm is not compatible with Multi Latent Attention (MLA) because QK norm requires full materialization of queries and keys during inference, which is not done in MLA. In this paper we suggest that controlling the changes to logits is important for stability. We show that these changes are controllable by assigning parameter-dependent learning rates to the query and key weights. We find that our cheap intervention allows us to increase the base learning rate of the network, outperform other methods in the MLA setting, and achieve performance competitive with QK norm when using Multi-head Attention.
Open AccessControlling changes to attention logits
Stability of neural network weights is critical when training transformer models. The query and key weights are particularly problematic, as they tend to grow large without any intervention. Applying normalization to queries and keys, known as `QK norm', fixes stability issues in practice, but is not always applicable. For example, QK norm is not compatible with Multi Latent Attention (MLA) because QK norm requires full materialization of queries and keys during inference, which is not done in MLA. In this paper we suggest that controlling the changes to logits is important for stability. We show that these changes are controllable by assigning parameter-dependent learning rates to the query and key weights. We find that our cheap intervention allows us to increase the base learning rate of the network, outperform other methods in the MLA setting, and achieve performance competitive with QK norm when using Multi-head Attention.
- Open Access
- 1 June 2025
- Published by American Meteorological Society in Monthly Weather Review
- Vol. 153 (6) , 1021-1033
How to Derive Skill from the Fractions Skill Score
The fractions skill score (FSS) is a widely used metric for assessing forecast skill, with applications ranging from precipitation to volcanic ash forecasts. By evaluating the fraction of grid squares exceeding a threshold in a neighborhood, the intuition is that it can avoid the pitfalls of pixelwise comparisons and identify length scales at which a forecast has skill. The FSS is typically interpreted relative to a “useful” criterion, where a forecast is considered skillful if its score exceeds a simple reference score. However, the typical reference score used is problematic, since it is not derived in a way that provides obvious meaning, does not scale with neighborhood size, and may not be exceeded by forecasts that have skill. We, therefore, provide a new method to determine forecast skill from the FSS, by deriving an expression for the FSS achieved by a random forecast, which provides a more robust and meaningful reference score to compare with. Through illustrative examples, we show that this new method considerably changes the length scales at which a forecast would be regarded as skillful and reveals subtleties in how the FSS should be interpreted. Significance Statement: Forecast verification metrics are crucial to assess accuracy and identify where forecasts can be improved. In this work, we investigate a popular verification metric, the fractions skill score, and derive a more robust method to decide if a forecast has sufficiently high skill. This new method significantly improves the quality of insights that can be drawn from this score.
- 1 June 2025
- Published by American Meteorological Society in Monthly Weather Review
- Vol. 153 (6) , 1021-1033
Open AccessHow to Derive Skill from the Fractions Skill Score
The fractions skill score (FSS) is a widely used metric for assessing forecast skill, with applications ranging from precipitation to volcanic ash forecasts. By evaluating the fraction of grid squares exceeding a threshold in a neighborhood, the intuition is that it can avoid the pitfalls of pixelwise comparisons and identify length scales at which a forecast has skill. The FSS is typically interpreted relative to a “useful” criterion, where a forecast is considered skillful if its score exceeds a simple reference score. However, the typical reference score used is problematic, since it is not derived in a way that provides obvious meaning, does not scale with neighborhood size, and may not be exceeded by forecasts that have skill. We, therefore, provide a new method to determine forecast skill from the FSS, by deriving an expression for the FSS achieved by a random forecast, which provides a more robust and meaningful reference score to compare with. Through illustrative examples, we show that this new method considerably changes the length scales at which a forecast would be regarded as skillful and reveals subtleties in how the FSS should be interpreted. Significance Statement: Forecast verification metrics are crucial to assess accuracy and identify where forecasts can be improved. In this work, we investigate a popular verification metric, the fractions skill score, and derive a more robust method to decide if a forecast has sufficiently high skill. This new method significantly improves the quality of insights that can be drawn from this score.
- 8 October 2024
- Published by MIT Press in Neural Computation
- Vol. 37 (1) , 1-32
Relating Human Error–Based Learning to Modern Deep RL Algorithms
In human error–based learning, the size and direction of a scalar error (i.e., the “directed error”) are used to update future actions. Modern deep reinforcement learning (RL) methods perform a similar operation but in terms of scalar rewards. Despite this similarity, the relationship between action updates of deep RL and human error–based learning has yet to be investigated. Here, we systematically compare the three major families of deep RL algorithms to human error–based learning. We show that all three deep RL approaches are qualitatively different from human error–based learning, as assessed by a mirror-reversal perturbation experiment. To bridge this gap, we developed an alternative deep RL algorithm inspired by human error–based learning, model-based deterministic policy gradients (MB-DPG). We showed that MB-DPG captures human error–based learning under mirror-reversal and rotational perturbations and that MB-DPG learns faster than canonical model-free algorithms on complex arm-based reaching tasks, while being more robust to (forward) model misspecification than model-based RL.
- 8 October 2024
- Published by MIT Press in Neural Computation
- Vol. 37 (1) , 1-32
Relating Human Error–Based Learning to Modern Deep RL Algorithms
In human error–based learning, the size and direction of a scalar error (i.e., the “directed error”) are used to update future actions. Modern deep reinforcement learning (RL) methods perform a similar operation but in terms of scalar rewards. Despite this similarity, the relationship between action updates of deep RL and human error–based learning has yet to be investigated. Here, we systematically compare the three major families of deep RL algorithms to human error–based learning. We show that all three deep RL approaches are qualitatively different from human error–based learning, as assessed by a mirror-reversal perturbation experiment. To bridge this gap, we developed an alternative deep RL algorithm inspired by human error–based learning, model-based deterministic policy gradients (MB-DPG). We showed that MB-DPG captures human error–based learning under mirror-reversal and rotational perturbations and that MB-DPG learns faster than canonical model-free algorithms on complex arm-based reaching tasks, while being more robust to (forward) model misspecification than model-based RL.
- Open Access
Stochastic Kernel Regularisation Improves Generalisation in Deep Kernel Machines
Recent work developed convolutional deep kernel machines, achieving 92.7% test accuracy on CIFAR-10 using a ResNet-inspired architecture, which is SOTA for kernel methods. However, this still lags behind neural networks, which easily achieve over 94% test accuracy with similar architectures. In this work we introduce several modifications to improve the convolutional deep kernel machine's generalisation, including stochastic kernel regularisation, which adds noise to the learned Gram matrices during training. The resulting model achieves 94.5% test accuracy on CIFAR-10. This finding has important theoretical and practical implications, as it demonstrates that the ability to perform well on complex tasks like image classification is not unique to neural networks. Instead, other approaches including deep kernel methods can achieve excellent performance on such tasks, as long as they have the capacity to learn representations from data.
Open AccessStochastic Kernel Regularisation Improves Generalisation in Deep Kernel Machines
Recent work developed convolutional deep kernel machines, achieving 92.7% test accuracy on CIFAR-10 using a ResNet-inspired architecture, which is SOTA for kernel methods. However, this still lags behind neural networks, which easily achieve over 94% test accuracy with similar architectures. In this work we introduce several modifications to improve the convolutional deep kernel machine's generalisation, including stochastic kernel regularisation, which adds noise to the learned Gram matrices during training. The resulting model achieves 94.5% test accuracy on CIFAR-10. This finding has important theoretical and practical implications, as it demonstrates that the ability to perform well on complex tasks like image classification is not unique to neural networks. Instead, other approaches including deep kernel methods can achieve excellent performance on such tasks, as long as they have the capacity to learn representations from data.
- 1 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (6) , 5719-5726
TouchSDF: A DeepSDF Approach for 3D Shape Reconstruction Using Vision-Based Tactile Sensing
Humans rely on their visual and tactile senses to develop a comprehensive 3D understanding of their physical environment. Recently, there has been a growing interest in exploring and manipulating objects using data-driven approaches that utilise high-resolution vision-based tactile sensors. However, 3D shape reconstruction using tactile sensing has lagged behind visual shape reconstruction because of limitations in existing techniques, including the inability to generalise over unseen shapes, the absence of real-world testing and limited expressive capacity imposed by discrete representations. To address these challenges, we propose TouchSDF, a Deep Learning approach for tactile 3D shape reconstruction that leverages the rich information provided by a vision-based tactile sensor and the expressivity of the implicit neural representation DeepSDF. This combination allows TouchSDF to reconstruct smooth and continuous 3D shapes from tactile inputs in simulation and real-world settings, opening up research avenues for robust 3D-aware representations and improved multimodal perception for robot manipulation.
- 1 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (6) , 5719-5726
TouchSDF: A DeepSDF Approach for 3D Shape Reconstruction Using Vision-Based Tactile Sensing
Humans rely on their visual and tactile senses to develop a comprehensive 3D understanding of their physical environment. Recently, there has been a growing interest in exploring and manipulating objects using data-driven approaches that utilise high-resolution vision-based tactile sensors. However, 3D shape reconstruction using tactile sensing has lagged behind visual shape reconstruction because of limitations in existing techniques, including the inability to generalise over unseen shapes, the absence of real-world testing and limited expressive capacity imposed by discrete representations. To address these challenges, we propose TouchSDF, a Deep Learning approach for tactile 3D shape reconstruction that leverages the rich information provided by a vision-based tactile sensor and the expressivity of the implicit neural representation DeepSDF. This combination allows TouchSDF to reconstruct smooth and continuous 3D shapes from tactile inputs in simulation and real-world settings, opening up research avenues for robust 3D-aware representations and improved multimodal perception for robot manipulation.
- Open Access
Batch size invariant Adam
We propose a batch size invariant version of Adam, for use in large-scale, distributed settings, in which the mini-batch is divided into micro-batches which are distributed among worker nodes. For the v term, standard Adam first computes the average over micro-batch gradients, then squares, while in the batch size invariant Adam proposed here, we first square the micro-batch gradients, then average. Previous work (e.g. Malladi et al. 2022) used an alternative approach that involved a square-root scaling of the learning rate, but this approach requires strong assumptions to work; in particular that the gradient variance dominates the square of the expected gradient. In contrast, the approach proposed here gives batch size invariance without this assumption. We confirm that in practice our scheme gives batch size invariance in a much larger range of scenarios than the previous approach.
Open AccessBatch size invariant Adam
We propose a batch size invariant version of Adam, for use in large-scale, distributed settings, in which the mini-batch is divided into micro-batches which are distributed among worker nodes. For the v term, standard Adam first computes the average over micro-batch gradients, then squares, while in the batch size invariant Adam proposed here, we first square the micro-batch gradients, then average. Previous work (e.g. Malladi et al. 2022) used an alternative approach that involved a square-root scaling of the learning rate, but this approach requires strong assumptions to work; in particular that the gradient variance dominates the square of the expected gradient. In contrast, the approach proposed here gives batch size invariance without this assumption. We confirm that in practice our scheme gives batch size invariance in a much larger range of scenarios than the previous approach.
- Open Access
Convolutional Deep Kernel Machines
Standard infinite-width limits of neural networks sacrifice the ability for intermediate layers to learn representations from data. Recent work (A theory of representation learning gives a deep generalisation of kernel methods, Yang et al. 2023) modified the Neural Network Gaussian Process (NNGP) limit of Bayesian neural networks so that representation learning is retained. Furthermore, they found that applying this modified limit to a deep Gaussian process gives a practical learning algorithm which they dubbed the deep kernel machine (DKM). However, they only considered the simplest possible setting: regression in small, fully connected networks with e.g. 10 input features. Here, we introduce convolutional deep kernel machines. This required us to develop a novel inter-domain inducing point approximation, as well as introducing and experimentally assessing a number of techniques not previously seen in DKMs, including analogues to batch normalisation, different likelihoods, and different types of top-layer. The resulting model trains in roughly 77 GPU hours, achieving around 99% test accuracy on MNIST, 72% on CIFAR-100, and 92.7% on CIFAR-10, which is SOTA for kernel methods.
Open AccessConvolutional Deep Kernel Machines
Standard infinite-width limits of neural networks sacrifice the ability for intermediate layers to learn representations from data. Recent work (A theory of representation learning gives a deep generalisation of kernel methods, Yang et al. 2023) modified the Neural Network Gaussian Process (NNGP) limit of Bayesian neural networks so that representation learning is retained. Furthermore, they found that applying this modified limit to a deep Gaussian process gives a practical learning algorithm which they dubbed the deep kernel machine (DKM). However, they only considered the simplest possible setting: regression in small, fully connected networks with e.g. 10 input features. Here, we introduce convolutional deep kernel machines. This required us to develop a novel inter-domain inducing point approximation, as well as introducing and experimentally assessing a number of techniques not previously seen in DKMs, including analogues to batch normalisation, different likelihoods, and different types of top-layer. The resulting model trains in roughly 77 GPU hours, achieving around 99% test accuracy on MNIST, 72% on CIFAR-100, and 92.7% on CIFAR-10, which is SOTA for kernel methods.
- Open Access
Bayesian Low-rank Adaptation for Large Language Models
Low-rank adaptation (LoRA) has emerged as a new paradigm for cost-efficient fine-tuning of large language models (LLMs). However, fine-tuned LLMs often become overconfident especially when fine-tuned on small datasets. Bayesian methods, with their inherent ability to estimate uncertainty, serve as potent tools to mitigate overconfidence and enhance calibration. In this work, we introduce Laplace-LoRA, which applies a Bayesian approach to the LoRA parameters. Specifically, Laplace-LoRA applies a Laplace approximation to the posterior over the LoRA parameters, considerably improving the calibration of fine-tuned LLMs.
Open AccessBayesian Low-rank Adaptation for Large Language Models
Low-rank adaptation (LoRA) has emerged as a new paradigm for cost-efficient fine-tuning of large language models (LLMs). However, fine-tuned LLMs often become overconfident especially when fine-tuned on small datasets. Bayesian methods, with their inherent ability to estimate uncertainty, serve as potent tools to mitigate overconfidence and enhance calibration. In this work, we introduce Laplace-LoRA, which applies a Bayesian approach to the LoRA parameters. Specifically, Laplace-LoRA applies a Laplace approximation to the posterior over the LoRA parameters, considerably improving the calibration of fine-tuned LLMs.
- Open Access
Taylor TD-learning
Many reinforcement learning approaches rely on temporal-difference (TD) learning to learn a critic. However, TD-learning updates can be high variance. Here, we introduce a model-based RL framework, Taylor TD, which reduces this variance in continuous state-action settings. Taylor TD uses a first-order Taylor series expansion of TD updates. This expansion allows Taylor TD to analytically integrate over stochasticity in the action-choice, and some stochasticity in the state distribution for the initial state and action of each TD update. We include theoretical and empirical evidence that Taylor TD updates are indeed lower variance than standard TD updates. Additionally, we show Taylor TD has the same stable learning guarantees as standard TD-learning with linear function approximation under a reasonable assumption. Next, we combine Taylor TD with the TD3 algorithm, forming TaTD3. We show TaTD3 performs as well, if not better, than several state-of-the art model-free and model-based baseline algorithms on a set of standard benchmark tasks.
Open AccessTaylor TD-learning
Many reinforcement learning approaches rely on temporal-difference (TD) learning to learn a critic. However, TD-learning updates can be high variance. Here, we introduce a model-based RL framework, Taylor TD, which reduces this variance in continuous state-action settings. Taylor TD uses a first-order Taylor series expansion of TD updates. This expansion allows Taylor TD to analytically integrate over stochasticity in the action-choice, and some stochasticity in the state distribution for the initial state and action of each TD update. We include theoretical and empirical evidence that Taylor TD updates are indeed lower variance than standard TD updates. Additionally, we show Taylor TD has the same stable learning guarantees as standard TD-learning with linear function approximation under a reasonable assumption. Next, we combine Taylor TD with the TD3 algorithm, forming TaTD3. We show TaTD3 performs as well, if not better, than several state-of-the art model-free and model-based baseline algorithms on a set of standard benchmark tasks.