Conditional Neural Processes

Abstract

Deep neural networks excel at function approximation, yet they are typically trained from scratch for each new function. On the other hand, Bayesian methods, such as Gaussian Processes (GPs), exploit prior knowledge to quickly infer the shape of a new function at test time. Yet GPs are computationally expensive, and it can be hard to design appropriate priors. In this paper we propose a family of neural models, Conditional Neural Processes (CNPs), that combine the benefits of both. CNPs are inspired by the flexibility of stochastic processes such as GPs, but are structured as neural networks and trained via gradient descent. CNPs make accurate predictions after observing only a handful of training data points, yet scale to complex functions and large datasets. We demonstrate the performance and versatility of the approach on a range of canonical machine learning tasks, including regression, classification and image completion.

References in corpus (15)

• Adam: A Method for Stochastic Optimization
• Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
• Prototypical Networks for Few-shot Learning
• Stochastic Backpropagation and Approximate Inference in Deep Generative Models
• Conditional Image Generation with PixelCNN Decoders
• Weight Uncertainty in Neural Networks
• DRAW: A Recurrent Neural Network For Image Generation
• Matching Networks for One Shot Learning
• Learning to reinforcement learn
• Bayesian Compression for Deep Learning
• Towards Deep Symbolic Reinforcement Learning
• Towards a Neural Statistician
• Few-shot Autoregressive Density Estimation: Towards Learning to Learn Distributions
• The Variational Homoencoder: Learning to learn high capacity generative models from few examples
• Fast Adaptation in Generative Models with Generative Matching Networks

Cited by in corpus (31)

• A Review of Uncertainty Quantification in Deep Learning: Techniques, Applications and Challenges
• Time Series Forecasting With Deep Learning: A Survey
• Deep Bayesian Inversion
• Meta-Learning without Memorization
• Imitation and Mirror Systems in Robots through Deep Modality Blending Networks
• Meta-Learning Mean Functions for Gaussian Processes
• Adaptive Deep Kernel Learning
• Selecting Relevant Features from a Multi-domain Representation for Few-shot Classification
• Verification of deep probabilistic models
• Improving predictions of Bayesian neural nets via local linearization
• Noise Contrastive Meta-Learning for Conditional Density Estimation using Kernel Mean Embeddings
• Deep Adaptive Design: Amortizing Sequential Bayesian Experimental Design
• Photoreverberation mapping of quasars in the context of LSST observing strategies
• Neural Process for Black-Box Model Optimization Under Bayesian Framework
• Towards Principled Uncertainty Estimation for Deep Neural Networks
• An empirical study of pretrained representations for few-shot classification
• Equivariant Learning of Stochastic Fields: Gaussian Processes and Steerable Conditional Neural Processes
• Function Contrastive Learning of Transferable Meta-Representations
• Graph Neural Processes: Towards Bayesian Graph Neural Networks
• Learning Partially Known Stochastic Dynamics with Empirical PAC Bayes
• Energy-Based Processes for Exchangeable Data
• How Tight Can PAC-Bayes be in the Small Data Regime?
• Contextual HyperNetworks for Novel Feature Adaptation
• Neural Likelihoods for Multi-Output Gaussian Processes
• Cubic Spline Smoothing Compensation for Irregularly Sampled Sequences
• Mini-Batch Consistent Slot Set Encoder for Scalable Set Encoding
• Subjectivity Learning Theory towards Artificial General Intelligence
• Viscos Flows: Variational Schur Conditional Sampling With Normalizing Flows
• Wasserstein Neural Processes
• Neural Approximation of an Auto-Regressive Process through Confidence Guided Sampling
• OR-Net: Pointwise Relational Inference for Data Completion under Partial Observation