recurrent | A redis-backed manager of recurrent jobs , for node.js | BPM library

I'm trying to learn Recurrent Neural Networks (RNN) with Flux.jl in Julia by following along some tutorials, like Char RNN from the FluxML/model-zoo.

I managed to build and train a model containing some RNN cells, but am failing to evaluate the model after training.

Can someone point out what I'm missing for this code to evaluate a simple (untrained) RNN?

I use Hangfire for scheduling jobs. In my setup I have defined two distinct queues. One is used to process recurrent jobs and jobs with a 'low' priority. The other queue is used for jobs that have a 'high' priority which are created ad hoc resulting from user input. Non recurrent jobs are scheduled like this to ensure they end up in the correct queue:

I'm recurrently running into a "problem" working with *ngIf. Often the function that returns boolean has a subscribe inside it. As the directive is checked multiple times, multiple requests end up being made and the page crashes. I really don't know how to solve it. Here's an example:

MyService.service.ts:

I used the forecast package to forecast the daily time-series of variable Y using its lag values and a time series of an external parameter X. I found nnetar model (a NARX model) was the best in terms of overall performance. However, I was not able to get the prediction of peaks of the time series well despite my various attempts with parameter tuning.

I then extracted the peak values (above a threshold) of Y (and of course this is not a regular time series anymore) and corresponding X values and tried to fit a regression model (note: not an autoregression model) using various models in carat package. I found out the prediction of peak values using brnn(Bidirectional recurrent neural networks) model just using X values is better than that of nnetar which uses both lag values and X values.

Now my question is how do I go from here to create ensamples of these two models (i.e whenever the prediction using brnn regression model ( or any other regression model) is better I want to replace the prediction using nnetar and move forward - I am mostly concerned about the peaks)? Is this a commonly used approach?

I'm trying to run the following code but I got an error. Did I miss something in the codes?

I am trying to create a sequential model with Tensorflow's (2.5) Keras API.
After training my model I've found out that I was not able to save my model because the config for Layer ModuleWrapper was not implemented, which brought a lot of confusion to me, because I was not using any Layer called 'ModuleWrapper'. I also did not use any self-made Layers.

After a lot of testing I figured out that the Keras Sequential API somehow does not recognize it's own Layers and replaces them with the Abstract Class(?) ModuleWrapper.

Any help on why this happens would be greatly appreciated!

I have an Umbraco 8 c# MVC website and have realized I'm having a lot of recurrent log warnings that are making my log file very big.

The log warning message is:

I'm astonished by the recurrent problem of size vector cannot be allocated in geom_raster function. This problem happens with small data set and in my case 1559 observations. I try to do:

Let's say I have a pytorch-model describing the evolution of some multidimensional system based on its own state x and an external actuator u. So x_(t+1) = f(x_t, u_t) with f being the artificial neural network from pytorch.

Now i want to solve a dynamic optimization problem to find an optimal sequence of u-values to minimize an objective that depends on x. Something like this:

min sum over all timesteps phi(x_t)

s.t.: x_(t+1) = f(x_t, u_t)

Additionally I also have some upper and lower bounds on some of the variables in x.

Is there an easy way to do this using a dynamic optimization toolbox like pyomo or gekko?

I already wrote some code that transforms a feedforward neural network to a numpy-function which can then be passed as a constraint to pyomo. The problem with this approach is, that it requires significant reprogramming-effort every time the structure of the neural network changes, so quick testing becomes difficult. Also integration of recurrent neural networks gets difficult because hidden cell states would have to be added as additional variables to the optimization problem.

I think a good solution could be to do the function evaluations and gradient calculations in torch and somehow pass the results to the dynamic optimizer. I'm just not sure how to do this.

Thanks a lot for your help!