Stochastic Intervention for Causal Effect Estimation

Duong, Tri Dung; Li, Qian; Xu, Guandong

doi:10.1109/ijcnn52387.2021.9533959

Cited by 5 publications

(4 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We provide a detailed discussion of study limitations in Supplementary section 8. Future work will address these by rendering images in simulations and using image-computable models, incorporating real eye-tracking and scene data in ROFL [80,106], testing our approach on more data from other brain areas such as MST [107,108], and using more sophisticated methods to measure representational alignment between ANNs and brains [109][110][111][112].…”

Section: Discussionmentioning

confidence: 99%

“…However, we note that this alignment score also has limitations, and for example, it is not a proper metric in the mathematical sense [109]. Future work will consider more sophisticated metrics for brain alignment [109][110][111][112].…”

Section: Study Limitations and Considerationsmentioning

confidence: 99%

“…In conclusion, refined and more comprehensive approaches are required to properly capture brain alignment [109][110][111][112].…”

Section: Brain Alignment Requires More Than Just Linear Regression Pe...mentioning

confidence: 99%

“…Another promising approach lies in focusing on the symmetry transformations inherent in the world [18,146]. By seeking representations that respect the underlying geometrical and group-theoretical structures of the world [147,148], and comparing them in principled ways [109][110][111], we may uncover new perspectives that provide valuable insights beyond the conventional and subjective notions of disentanglement.…”

Section: Disentanglement Is In the Eyes Of The Beholdermentioning

confidence: 99%

See 3 more Smart Citations

Hierarchical VAEs provide a normative account of motion processing in the primate brain

Vafaii,

Yates,

Butts

2023

Preprint

View full text Add to dashboard Cite

The relationship between perception and inference, as postulated by Helmholtz in the 19th century, is paralleled in modern machine learning by generative models like Variational Autoencoders (VAEs) and their hierarchical variants. Here, we evaluate the role of hierarchical inference and its alignment with brain function in the domain of motion perception. We introduce a novel synthetic data framework, Retinal Optic Flow Learning (ROFL), which enables control over motion statistics and their causes. We introduce a new hierarchical VAE and test it against alternative models on two downstream tasks: (i) predicting ground truth causes of retinal optic flow (e.g., self-motion); and (ii) predicting the responses of neurons in the motion processing pathway of primates. We manipulate the model architectures (hierarchical versus non-hierarchical), loss functions, and the causal structure of the motion stimuli. We found that hierarchical latent structure in the model leads to several improvements. First, it improves the linear decodability of ground truth variables and does so in a sparse and disentangled manner. Second, our hierarchical VAE outperforms previous state-of-the-art models in predicting neuronal responses with a performance gain of over 2x and exhibits sparse latent-to-neuron relationships. Finally, these results depend on the causal structure of the world, indicating that alignment between brains and artificial neural networks depends not only on architecture but also on matching ecologically relevant stimulus statistics. Collectively, these results support the notion that hierarchical Bayesian inference underlies the brain’s understanding of the world, and hierarchical VAEs offer an effective means of capturing this understanding.

show abstract