We study Bayesian hypernetworks: a framework for approximate Bayesian inference in neural networks. A Bayesian hypernetwork $\h$ is a neural network which learns to transform a simple noise distribution, $p(\vecε) = \N(\vec 0,\mat I)$, to a distribution $q(\pp) := q(h(\vecε))$ over the parameters $\pp$ of another neura…
We compress large neural networks for quick adaptation to specific contexts.
problem How to quickly adapt a pretrained large neural network to specific contexts.
method Propose a Bayesian hypernetwork framework to compress the network and encourage sparsity.
result Generated compressed networks are significantly smaller than baseline methods.
Wide hypernetworks don't guarantee convergence under gradient descent.
problem Theoretical guarantees for wide hypernetworks in over-parameterized settings.
method Analyzing infinitely wide hypernetworks and their convergence properties.
result Infinitely wide hypernetworks do not guarantee convergence to a global minimum under gradient descent.
This work explores maximum likelihood optimization of neural networks through hypernetworks. A hypernetwork initializes the weights of another network, which in turn can be employed for typical functional tasks such as regression and classification. We optimize hypernetworks to directly maximize the conditional likelih…
Hypernetworks are simplified simplicial complexes with curvature.
problem Representing hypernetworks geometrically for analysis.
method Hypernetworks are interpreted as posets, which are simplicial complexes with Forman Ricci curvature.
result Hypernetworks have intrinsic curvature that correlates with their Euler characteristic.
Artificial neural networks suffer from catastrophic forgetting when they are sequentially trained on multiple tasks. To overcome this problem, we present a novel approach based on task-conditioned hypernetworks, i.e., networks that generate the weights of a target model based on task identity. Continual learning (CL) i…
Improved hypernetwork for efficient neural network hyperparameter tuning.
problem Efficiently optimizing hyperparameters in neural networks.
method Proposed Δ-STN architecture focusing on accurate best-response Jacobian approximation. result Significantly improved hyperparameter tuning accuracy and stability.
A new algorithm optimizes multiple molecular properties efficiently.
problem Designing molecules with conflicting objectives and costly evaluations.
method Multi-objective Bayesian optimization with GFlowNets.
result HN-GFN samples diverse molecules from an approximate Pareto front.
Graph neural networks are currently leading the performance charts in learning-based molecule property prediction and classification. Computational chemistry has, therefore, become the a prominent testbed for generic graph neural networks, as well as for specialized message passing methods. In this work, we demonstrate…
This work introduces an integrative approach based on Q-analysis with machine learning. The new approach, called Neural Hypernetwork, has been applied to a case study of pulmonary embolism diagnosis. The objective of the application of neural hyper-network to pulmonary embolism (PE) is to improve diagnose for reducing …
Generative models can be unfair and unstable; new methods improve fairness and stability.
problem Generative models unfairly penalize minority data and suffer from MADness.
method Intentionally designed hypernetworks and regularization terms.
result Generative models are more fair, stable, and unbiased with new methods.
Unified framework for learning function representations using INRs and Transformers.
problem Scalability and efficiency limitations in existing generative models.
method Integrates INRs and Transformer-based hypernetworks into latent variable models.
result Improved scalability, expressiveness, and generalization over existing models.
Quantum computing improves training of binary neural networks.
problem Training binary neural networks (BiNNs) is challenging.
method Employing a Variational Quantum Algorithm to generate binary weights using quantum circuit measurements.
result The proposed methods improve trainability and generalization of BiNNs.
In the context of learning to map an input I to a function hI:X→R, two alternative methods are compared: (i) an embedding-based method, which learns a fixed function in which I is encoded as a conditioning signal e(I) and the learned function takes the form hI(x)=q(x,e(I)), and (ii) …
HKF uses neural networks to adapt Kalman filters for dynamic channel tracking.
problem Tracking channels with varying dynamics and Doppler values.
method Combines Kalman filters with hypernetworks for dynamic adaptation.
result HKF achieves up to 2dB gain over Kalman filters at high Doppler values.
Single training run learns optimal VAE parameters for various β values.
problem Training VAEs with varying β values for optimal trade-off between distortion and rate.
method Introduced Multi-Rate VAE (MR-VAE) using hypernetworks to map β to optimal parameters.
result MR-VAEs can construct the full rate-distortion curve without additional training.
Differentially private learning avoids iterative optimization in parameter space.
problem Differentially private training of neural networks
method Hypernetworks trained on public datasets
result Significantly reduces noise in parameter space
Neural architecture search (NAS) automatically finds the best task-specific neural network topology, outperforming many manual architecture designs. However, it can be prohibitively expensive as the search requires training thousands of different networks, while each can last for hours. In this work, we propose the Gra…
Modern neural networks tend to be overconfident on unseen, noisy or incorrectly labelled data and do not produce meaningful uncertainty measures. Bayesian deep learning aims to address this shortcoming with variational approximations (such as Bayes by Backprop or Multiplicative Normalising Flows). However, current appr…
Recent work on mode connectivity in the loss landscape of deep neural networks has demonstrated that the locus of (sub-)optimal weight vectors lies on continuous paths. In this work, we train a neural network that serves as a hypernetwork, mapping a latent vector into high-performance (low-loss) weight vectors, general…
Proposes a framework for semi-supervised continual learning from sequentially arriving data.
problem Learning from data with changing task distribution over time, especially in domains with a mix of labeled and unlabeled data.
method Meta-Consolidation for Continual Semi-Supervised Learning (MCSSL) framework with a hypernetwork and semi-supervised auxiliary classifier.
result Significant improvements in continual semi-supervised learning setting.
FDN improves probabilistic regressors' adaptability to distribution shifts.
problem Overconfidence in modern probabilistic regressors under distribution shift.
method FDN uses input-conditioned distributions over network weights, trained with a Monte Carlo beta-ELBO objective.
result FDN produces predictive mixtures whose dispersion adapts to the input, providing shift-aware uncertainty.
This paper introduces a new formulation of the Conic Gromov-Wasserstein distance for comparing complex network structures.
problem Comparing measures of unequal mass and complex network structures.
method Novel semi-coupling formulation and extension to hypernetworks.
result Establishes fundamental properties and robustness of CGW metric.
Hypernetworks are neural networks that generate weights for another neural network. We formulate the hypernetwork training objective as a compromise between accuracy and diversity, where the diversity takes into account trivial symmetry transformations of the target network. We explain how this simple formulation gener…
TensorHyper-VQC improves VQC scalability and robustness.
problem Scalability and noise sensitivity in VQC.
method Tensor-train-guided hypernetwork framework.
result TensorHyper-VQC achieves superior performance and robust noise tolerance.
In this article we show the duality between tensor networks and undirected graphical models with discrete variables. We study tensor networks on hypergraphs, which we call tensor hypernetworks. We show that the tensor hypernetwork on a hypergraph exactly corresponds to the graphical model given by the dual hypergraph. …
SVH-PSL uses Stein Variational Gradient Descent and Hypernetworks to improve Pareto set learning for expensive MOO.
problem Fragmented surrogate models and pseudo-local optima in expensive multi-objective optimization problems.
method SVH-PSL integrates Stein Variational Gradient Descent (SVGD) with Hypernetworks to address fragmentation and pseudo-local optima.
result SVH-PSL significantly improves the quality of the learned Pareto set, offering a promising solution for expensive MOO.
Classifying human cognitive states from behavioral and physiological signals is a challenging problem with important applications in robotics. The problem is challenging due to the data variability among individual users, and sensor artefacts. In this work, we propose an end-to-end framework for real-time cognitive wor…
A new framework enables real-time task trade-off control.
problem Conflict between multiple related tasks in a fixed model capacity.
method Formulates MTL as a preference-conditioned multiobjective optimization problem; uses a hypernetwork-based neural network.
result A single model can handle different trade-off preferences among multiple tasks.
Motivated by the human way of memorizing images we introduce their functional representation, where an image is represented by a neural network. For this purpose, we construct a hypernetwork which takes an image and returns weights to the target network, which maps point from the plane (representing positions of the pi…
This paper won 1st place in forecasting and investment challenges, improving on meta-learning and parametric models.
problem Forecasting and investment challenges in time-series data.
method Hypernetworks and adversarial portfolios to design time-series models.
result Outperformed state-of-the-art meta-learning methods and conventional parametric models.
LatentTrack generates model parameters online for nonstationary data.
problem Online probabilistic prediction under nonstationary dynamics.
method Sequential neural architecture with latent filtering and amortized inference.
result Consistently lower negative log-likelihood and mean squared error than baselines.
Knowledge graphs are graphical representations of large databases of facts, which typically suffer from incompleteness. Inferring missing relations (links) between entities (nodes) is the task of link prediction. A recent state-of-the-art approach to link prediction, ConvE, implements a convolutional neural network to …
The manual design of analog circuits is a tedious task of parameter tuning that requires hours of work by human experts. In this work, we make a significant step towards a fully automatic design method that is based on deep learning. The method selects the components and their configuration, as well as their numerical …
Generative neural nets learn deep policies conditioned on goals.
problem Learning optimal policies for specific goals in reinforcement learning.
method Goal-conditioned neural nets that generate deep neural policies.
result Single learned policy generator can achieve any desired return.
A scalable framework uses Langevin sampling to approximate neural network models of evolving processes.
problem Uncertainty quantification in neural network models of dynamic systems.
method Flexible data model based on NODE, joint learning of data model and posterior parameters, Langevin sampling.
result Demonstrated performance on chemical reaction and material physics data, compared favorably to variational inference.
A new method lifts training of input-convex neural networks to avoid dead weights and plateaued loss.
problem Training input-convex neural networks with non-negative weights.
method Introduces a hypernetwork that emits non-negative weights from a summary of the input batch, adding stochasticity to soften the loss landscape.
result The lift method achieves lower test loss than projected gradient descent and direct softplus reparametrization.
Hypermodels improve exploration efficiency and accuracy.
problem Efficiently approximating Thompson sampling with large ensembles.
method Introducing hypermodels as a generalization of ensembles, including linear and neural network hypermodels.
result Hypermodels enable more accurate exploration and performance gains over Thompson sampling.
Spatio-temporal (ST) data, which represent multiple time series data corresponding to different spatial locations, are ubiquitous in real-world dynamic systems, such as air quality readings. Forecasting over ST data is of great importance but challenging as it is affected by many complex factors, including spatial char…
Paper proposes a method to control robots of different shapes efficiently.
problem Learning optimal control policies for robots of various shapes is challenging.
method Hierarchical architecture with hypernetworks and fixed attention mechanism.
result Method improves learning performance and generalizes to unseen morphologies.
CoDA adapts dynamics models to new physical systems by conditioning on context.
problem Generalizing to new physical systems with shared dynamics but different contexts.
method Context-informed dynamics adaptation (CoDA) using multiple environments and a hypernetwork.
result State-of-the-art generalization results on nonlinear dynamics.
DISCO predicts system states from short trajectories using an evolved operator.
problem Predicting next states of dynamical systems governed by unknown PDEs.
method DISCO uses a hypernetwork to generate parameters of a smaller operator network for state prediction.
result DISCO achieves state-of-the-art performance with fewer training epochs and generalizes well.
New MTPP model offers interpretable predictions with state-of-the-art performance.
problem Inexpressive models lack interpretability, while neural models sacrifice interpretability for performance.
method Extends Hawkes process to a hypernetwork with a latent space, making it flexible and interpretable.
result Achieves state-of-the-art performance across various tasks and metrics.
We simplify Volterra process predictions by reducing dimensionality and using a tailored deep learning model.
problem Predicting the conditional law of Volterra processes with stochastic volatility is challenging due to high dimensionality and non-smoothness.
method We developed a stable dimension reduction technique onto a low-dimensional statistical manifold of non-positive curvature and introduced a sequentially deep learning model tailored to this geometry.
result Our model can approximate the conditional law of Volterra processes with approximation rates achievable only with very large networks.
Novel NAS method balances performance and hardware metrics efficiently.
problem Challenging multi-objective optimization in neural architecture search.
method Parameterizes joint architectural distribution via hypernetwork conditioned on hardware features and preferences.
result Zero-shot transferability to new devices with representative and diverse architectures.
Study evaluates CL methods in RNNs, highlighting differences from feedforward networks.
problem Preventing catastrophic forgetting in RNNs processing sequential data.
method Comprehensive evaluation of CL methods, including elastic weight consolidation and hypernetworks.
result Weight-importance methods perform similarly regardless of sequence length but require more stability for high working memory demands.
Neural decoders were shown to outperform classical message passing techniques for short BCH codes. In this work, we extend these results to much larger families of algebraic block codes, by performing message passing with graph neural networks. The parameters of the sub-network at each variable-node in the Tanner graph…
Neural framework for conditional OT maps learns from categorical and continuous variables.
problem Learning conditional optimal transport maps between complex distributions.
method Hypernetwork generates adaptive transport layer parameters based on conditioning variables.
result Our method outperforms simpler conditioning methods in comprehensive ablation studies.