A Deep Generative Model Imitating Predictive Coding in Human Brain
In recent years, the development of deep generative models for prediction has been attracting attention. In our study, we focus on predictive coding, a concept from the neuroscience literature that hypothesizes the brain is constantly making predictions of sensory input, and attempt to develop a human brain-like prediction model. PredNet is a deep learning model based on the concept of predictive coding, however, it can predict the next image with the same prediction interval, but not with any intervals. On the other hand, Temporal Differential Variational Auto-Encoder (TD-VAE) can predict the next images with any prediction intervals, although it is not a model reflecting human brain function. In this work, we develop a new human brain-like prediction model by unifying PredNet and TD- VAE, combining both predictive coding and flexible interval prediction abilities in one single model. Through experiments on the KITTI Vision Benchmark, we confirmed that our proposed model can predict the next images correctly with flexible prediction intervals. We also investigated the correlation between the feature values of representation layers in the model architecture and human brain activity data evoked under natural video stimulation.
Code (0)
등록된 구현이 없습니다.
Tasks
PredictionPrediction IntervalsSimilar Papers 제목 키워드 기반
Decoding Predictive Inference in Visual Language Processing via Spatiotemporal Neural Coherence
Human language processing relies on the brain's capacity for predictive inference. We present a machine learning framework for decoding neural (EEG) responses to dynamic visual language stimuli in Deaf signers. Using coh…
Language Reconstruction with Brain Predictive Coding from fMRI Data
Many recent studies have shown that the perception of speech can be decoded from brain signals and subsequently reconstructed as continuous language. However, there is a lack of neurological basis for how the semantic in…
Predictive Coding: a Theoretical and Experimental Review
Predictive coding offers a potentially unifying account of cortical function -- postulating that the core function of the brain is to minimize prediction errors with respect to a generative model of the world. The theory…
A Deep Predictive Coding Network for Learning Latent Representations
It has been argued that the brain is a prediction machine that continuously learns how to make better predictions about the stimuli received from the external environment. For this purpose, it builds a model of the world…
Predictive Coding Theories of Cortical Function
Predictive coding is a unifying framework for understanding perception, action and neocortical organization. In predictive coding, different areas of the neocortex implement a hierarchical generative model of the world t…
Bayesian Inference