2021 Fiscal Year Research-status Report
Experimental and theoretical investigation of Bayesian inference across multiple layers in somatosensory cortex
| Project/Area Number |
19K20348
|
|---|
| Research Institution | Okinawa Institute of Science and Technology Graduate University |
|---|
Principal Investigator |
LI Yuzhe 沖縄科学技術大学院大学, 神経計算ユニット, スタッフサイエンティスト (30815436)
|
|---|
| Project Period (FY) |
2019-04-01 – 2023-03-31
|
| Keywords | calcium imaging / factor analysis / Bayesian PCA / dimensionality reduction |
|---|
| Outline of Annual Research Achievements |
We started to analyze our calcium imaging data using factor analysis. The goal of factor analysis is to reduce the dimension of the neural recordings and use the reduced dimensions to find the relation between neural activities and stimulus.
While applying factor analysis to calcium imaging data, we found that the existing methods could not sufficiently reduce the dimensionality of the fluorescence recordings. Therefore, we proposed a new method based on Bayesian PCA by adding an extra Bayesian formation to the latent variables.
Our new method showed good reconstruction performance using the dimension-reduced latent variables. and application on our data showed the low-dimensional latent variable captured the motion features during locomotion and the directions of the auditory stimulus.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
Because of the pandemic, the animal experiments had to be postponed, but during this period, we analyzed the resting-state dataset and developed a new method for factor analysis on calcium imaging data. We also investigated the relations among neurons from connectivity, causality, clustering, and network features.
|
| Strategy for Future Research Activity |
Although we developed a new method for factor analysis on calcium imaging data and analyzed the resting data from connectivity, causality, and network features, our goal of the project is to investigate the Bayesian feature of neural activity in predicting the future stimulus input. Therefore, we need to resume animal experiments and collect more data for the Bayesian modeling of neural activities. Moreover, the Bayesian modeling of the multi-layer cortical activities needs to be done.
|
| Causes of Carryover |
1. We need to purchase new mice and prism lens for the following experiments.
2. Some consumable parts of the setup need to be replaced.
3. Cost of submitting our papers
|
