↓Skip to main content

🏢 University of Cambridge

Generating Origin-Destination Matrices in Neural Spatial Interaction Models
·2770 words·14 mins· loading · loading
AI Applications Smart Cities 🏢 University of Cambridge
GeNSIT: a neural framework efficiently generates origin-destination matrices for agent-based models, outperforming existing methods in accuracy and scalability by directly operating on the discrete sp…
Fearless Stochasticity in Expectation Propagation
·1730 words·9 mins· loading · loading
🏢 University of Cambridge
This paper introduces EP-η and EP-μ, novel EP variants remarkably robust to Monte Carlo noise, achieving improved speed and accuracy.
End-to-End Ontology Learning with Large Language Models
·6489 words·31 mins· loading · loading
AI Generated Natural Language Processing Large Language Models 🏢 University of Cambridge
OLLM: An end-to-end LLM method builds ontologies from scratch, outperforming subtask approaches and improving semantic accuracy with novel evaluation metrics.
Efficient Lifelong Model Evaluation in an Era of Rapid Progress
·2830 words·14 mins· loading · loading
AI Generated Computer Vision Image Classification 🏢 University of Cambridge
Sort & Search: 1000x faster lifelong model evaluation!
Deep Learning Through A Telescoping Lens: A Simple Model Provides Empirical Insights On Grokking, Gradient Boosting & Beyond
·3053 words·15 mins· loading · loading
AI Generated Machine Learning Deep Learning 🏢 University of Cambridge
A simple, yet accurate model unveils deep learning’s mysteries, providing empirical insights into grokking, double descent, and gradient boosting, offering a new lens for analyzing neural network beha…
Deep Equilibrium Algorithmic Reasoning
·2322 words·11 mins· loading · loading
Machine Learning Deep Learning 🏢 University of Cambridge
Deep Equilibrium Algorithmic Reasoners (DEARs) achieve superior performance on algorithmic tasks by directly solving for the equilibrium point of a neural network, eliminating the need for iterative r…
Decomposable Transformer Point Processes
·2120 words·10 mins· loading · loading
AI Generated Machine Learning Deep Learning 🏢 University of Cambridge
Decomposable Transformer Point Processes (DTPP) dramatically accelerates marked point process inference by using a mixture of log-normals for inter-event times and Transformers for marks, outperformin…
Data-Driven Discovery of Dynamical Systems in Pharmacology using Large Language Models
·1652 words·8 mins· loading · loading
AI Applications Healthcare 🏢 University of Cambridge
LLMs iteratively discover and refine interpretable dynamical systems models, achieving high accuracy and uncovering new insights; demonstrated by a novel Warfarin model.
Context-Aware Testing: A New Paradigm for Model Testing with Large Language Models
·4422 words·21 mins· loading · loading
Machine Learning Large Language Models 🏢 University of Cambridge
Context-Aware Testing (CAT) revolutionizes ML model testing by using contextual information to identify relevant failures, surpassing traditional data-only methods.
Constructing Semantics-Aware Adversarial Examples with Probabilistic Perspective
·1825 words·9 mins· loading · loading
Computer Vision Image Classification 🏢 University of Cambridge
Researchers developed semantics-aware adversarial examples using a probabilistic approach, achieving higher success rates in bypassing defenses while remaining undetectable to humans.
CLUES: Collaborative Private-domain High-quality Data Selection for LLMs via Training Dynamics
·2368 words·12 mins· loading · loading
AI Generated Natural Language Processing Large Language Models 🏢 University of Cambridge
CLUES: Collaborative learning selects high-quality private data for LLM fine-tuning via training dynamics, significantly boosting performance in diverse domains.
Bias in Motion: Theoretical Insights into the Dynamics of Bias in SGD Training
·2198 words·11 mins· loading · loading
AI Theory Fairness 🏢 University of Cambridge
AI systems acquire bias during training, impacting accuracy across sub-populations. This research unveils bias’s dynamic nature, revealing how classifier preferences shift over time, influenced by dat…
Beyond Slow Signs in High-fidelity Model Extraction
·2693 words·13 mins· loading · loading
AI Generated Machine Learning Deep Learning 🏢 University of Cambridge
Researchers drastically sped up high-fidelity deep learning model extraction, improving efficiency by up to 14.8x and challenging previous assumptions on the extraction bottleneck.
Automatically Learning Hybrid Digital Twins of Dynamical Systems
·2680 words·13 mins· loading · loading
AI Applications Healthcare 🏢 University of Cambridge
AI autonomously designs highly effective hybrid digital twins by combining neural networks and mechanistic models, significantly advancing digital twin technology.
Approximately Equivariant Neural Processes
·2389 words·12 mins· loading · loading
Machine Learning Deep Learning 🏢 University of Cambridge
Boosting meta-learning, this paper introduces a novel, flexible approach to create approximately equivariant neural processes that outperform both non-equivariant and strictly equivariant counterparts…
An Improved Empirical Fisher Approximation for Natural Gradient Descent
·2632 words·13 mins· loading · loading
Machine Learning Optimization 🏢 University of Cambridge
Improved Empirical Fisher (iEF) approximation significantly boosts the performance of Natural Gradient Descent (NGD) optimizers, offering superior convergence and generalization.
Accelerating Relative Entropy Coding with Space Partitioning
·1881 words·9 mins· loading · loading
Machine Learning Deep Learning 🏢 University of Cambridge
Space partitioning dramatically speeds up relative entropy coding (REC) for neural compression, achieving 5-15% better bitrates than previous methods.
A theoretical design of concept sets: improving the predictability of concept bottleneck models
·1648 words·8 mins· loading · loading
AI Theory Interpretability 🏢 University of Cambridge
Boosting concept bottleneck model predictability, this paper introduces a theoretical framework linking concept set properties to model performance, proposing a method for effective concept identifica…
A Generative Model of Symmetry Transformations
·3610 words·17 mins· loading · loading
Machine Learning Generative Learning 🏢 University of Cambridge
Generative model learns data symmetries for improved efficiency and higher test log-likelihoods.