A cheap/free high memory server for machine learning algorithms research (open source)?

I’m trying to fit some kind of causal model to continuous value data by solving differential equations probabilistically (machine learning).

Currently I’m solving complex-valued vector quadratic differential equation so there are more cross correlations between variables.

dx(t)/dt = diag(Ax(t)x(t)^h) + Bx(t) + c + f(t)

or just

dx(t)/dt = diag(Ax(t)x(t)^h) + Bx(t) + c

diag() takes diagonal of the square matrix.

But my diff. eq. math is rusty because I have studied differential equations 20 years ago. I solved the equation in 1-dimensional case but would need help for vector valued x(t).

Would someone point me to appropriate material?

EDIT: I did edit the question to be a bit more clear to read.

Dear biostats community,

I am trying to build a Cox (Proportional Hazards) Regression and have a dataset with several variables. I am trying to decide which variables are the most useful to use as covariates in my model. Which method do you use and recommend for doing this? I thought at first to do a univariate analysis and see which variables don't have a significant survival difference to exclude them but as I understand such procedures have the issue that they don't take into account the interaction between the different variables.

Thank you very much!!

Gabriel

How do we determine the appropriate number of hidden layers for the problem so that it positively affects the solution of the desired problem?

How do we understand the impact of adding an additional hidden layer on improving workflow? Where exactly will the improvement be?

I'd like to report fold-change values as a percentage change compared to the control group (fold-change = 1)... Would my reasoning be correct for the three examples?

1) fold-change 1,65 -> up-regulation by 65% or up-regulation to 165% compared to the control group

2) fold-change 7,08 -> up-regulation by 608% or up-regulation to 708% compared to the control group

3) fold-change 0,14 -> down-regulation by 86% or down-regulation to 14% compared to the control group

Thank you all for your help!

How is the depth and width of a neural network in machine learning chosen for optimal processing?

Given that neural networks are tending to be deeper now (thanks to more efficient coding and faster machines) there's greater variability in the number of parameters to choose from. However, in spite of this 'spoilt for choice' scenario, there must be general guidelines on the number of layers and the widths of these layers in a neural network, so networks can be trained more rapidly, data is evaluated more quickly and computer resources are not wasted.

Input data parameters are normally fixed, such as the number of input nodes and the dynamic range in the data. Output parameters are also fixed, such as the number of output nodes and the quality of the data. Furthermore, the complexity of the neural network goes linearly with the depth of the network, but as the square of the width of the network.

Since the above 'boundary conditions' are generally known for an application of neural networks, how does one go about choosing the width and depth of a neural network?

many thanks,

neil

I am interested in publishing a recent scientific study (brief report) that may have some political ramifications. I am searching for a non-peer reviewed journal or publication platform. This study will benefit the general public as much as the scientific community.

I trained artificial neural network with 11 inputs and one output. There are total 74 data. I need to compare the predicted values with values calculated from equations of design codes. So can someone help me how can I get the predicted output

Hi all,

I hope you are doing well.

Currently, I am working on my university project and my topic is Stability of Risk Preferences. I am planning to use the available longitudinal dataset to estimate the effect of several regressors on the regressand over the period of various years. I think via simple FE/RE estimation I can measure the changes in risk preferences over time and see if they are (in)stable. But I am not sure as I do not know any other techniques for this approach. I wanted to get the thought of experts if I am in the right direction as I am quite new in this field. Any advice will be highly appreciated.

Thanks in advance.

Hello,

I was hoping to reach out to this research community to seek some perspective and advice on a statistical task I am working on through R. The objective is very simple and straightforward, but I am taken back by how complex the process may actually be.

The objective is outlined as follows: I have three boxes in a single room. One is blue, one is green, and one is red. In each box is a temperature reader recording a single temperature reading per day over a 30-day period. So now for each box, I have data that contains a single temperature reading for each day for 30 days.

To my understanding, this would be time-series data. Now I want to address the question: Is there a significant difference in temperature between the three boxes? Phrased differently: Does the choice of color dictate the overall temperature in the box?

Of course I can just take the mean temperature for the 30-day period for each box and just compare that, but this doesn't seem complete. Since I am working with categorical data (color of box) and continuous data (temperature), I was thinking I need to perform an ANOVA test. Then I would perform the Tukey HSD post-hoc test to look for individual comparisons, such that in addition to comparing all the boxes together, I would compare the blue box to the red box and the blue box to the green box, and compare the green box to the red box. This however, would just be looking at the affect of color on temperature, ignoring the whole time-series component. How can I add the time component to this?

I know R has the time series function: ts(), so would it work to just make each data set for each box a time-series abject, and then just run the ANOVA and Tukey HSD post-hoc test on these time-series objects? How should I best proceed with my objective? I know there are factors such as seasonality and auto-correlation here, but I am not sure how to incorporate these considerations. Is there a simple way to do all of this? Could you perhaps provide some R code examples?

Thank you so much!