Inferring Causal Directions in Errors-in-Variables Models

Authors: Yulai Zhang, Guiming Luo

AAAI 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Some experiments are included to illustrate our method.
Researcher Affiliation Academia Yulai Zhang and Guiming Luo School of Software, Tsinghua University Beijing, 100084, China P.R. {zhangyl08@mails, gluo@mail}.tsinghua.edu.cn
Pseudocode Yes We propose an EM style algorithm in this work to solve (6) and (8) together as follow: Step 1. Select the model order n and the length of the instrumental variable q. Initialize the values of the optimization variables in (6) and (8). Step 2. Solve (6), get the result of θ(t) at tth step. Step 3. Solve (8), and get σ2 ex(t). While |σ2 ex(t)/σ2 ex(t 1) 1| > δ, go back to Step 2. Step 4. Calculate the pdf px(x) using (7) Step 5. Calculate ρxy by (5) Step 6. Exchange x and y, do Step 2 to Step 5 again to calculate ρyx. Step 7. If |ρxy| < |ρyx|, the causal direction is x causes y, and if |ρyx| < |ρxy|, the causal direction is y causes x.
Open Source Code No No statement regarding the release of source code or a link to a code repository was found.
Open Datasets No The paper describes generating synthetic data: 'The data is from second order polynomial models. θd is randomly generated between [0,2]. The input data x is generated by a uniform distributed between -0.5 and 1.5. σ2 ex and σ2 ey are equally selected from 0.1 to 0.3.' It does not refer to a publicly available dataset with concrete access information.
Dataset Splits No The paper does not specify concrete dataset splits (e.g., percentages or counts) for training, validation, or testing.
Hardware Specification No No specific hardware details (e.g., CPU, GPU models, memory) used for running experiments were mentioned in the paper.
Software Dependencies No No specific software dependencies with version numbers (e.g., libraries, frameworks, or solvers with their exact versions) were mentioned in the paper.
Experiment Setup Yes The paper describes experimental parameters such as the generation of input data x 'by a uniform distributed between -0.5 and 1.5', noise variances 'σ2 ex and σ2 ey are equally selected from 0.1 to 0.3', and parameters for the EM-style algorithm: 'Select the model order n and the length of the instrumental variable q. Initialize the values of the optimization variables in (6) and (8)... While |σ2 ex(t)/σ2 ex(t 1) 1| > δ'.