Comments (5)
Set optimizer=None in https://github.com/probml/ssm-jax/blob/main/ssm_jax/hmm/models/bernoulli_hmm.py#L68
from dynamax.
Here are the details from my book 1. Note that the add one smoothng prior corresponds to Beta(2,2) which is much stronger than your current Beta(1.1, 1.1). This might explain your numerical problems.
from dynamax.
Replace beta.mode in https://github.com/probml/ssm-jax/blob/main/ssm_jax/hmm/models/bernoulli_hmm.py#L99 with pseudo counts
I actually like the Beta(count1, count0).mode()
formulation! I think it shows off the general form of the M-steps.
from dynamax.
I think we've all settled on using the Beta(count1, count0).mode()
. Could I close this issue?
from dynamax.
Yup!
from dynamax.
Related Issues (20)
- Simplify initial elem in LGSSM parallel inference code. HOT 4
- Is there a planned implementation of a hierarchical hidden markov model?
- impose constraints on the transition probability.
- pass inputs into the LDS model HOT 3
- How to`fit_constrained` and handle missing values HOT 3
- Add support for diagonal emissions noise in LGSSM models
- `psd_solve` leads to numerical errors
- Support for rSLDS? HOT 1
- Add support for parallel hmm sampler HOT 5
- GLM-HMM HOT 4
- Fit CategoricalHMM with available data? HOT 1
- Issue with LinearGaussiamSMM with time-varying emissions weights HOT 1
- Example of EKF with an input? HOT 3
- m_step for LinearRegressionHMMEmissions and DiagonalGaussianHMMEmissions does not check whether parameters are trainable
- NAN returned in large data and large iteration number HOT 1
- Order of state vector in Kalman Filter example docs/notebook incorrect HOT 1
- dynamax is weirdly slow in my HMM benchmark
- Question about hierarchical model support in dynamax
- Add general info form message passing algorithms for linear Gaussian SSMs
- dynmax glm-hmm using jagged input arrays with differing E and M steps HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from dynamax.