Causality Preserving Chaotic Transformation and Classification using Neurochaos Learning
Authors: Harikrishnan N B, Aditi Kathpalia, Nithin Nagaraj
NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this work, a recently proposed brain inspired learning algorithm namely-Neurochaos Learning (NL) is used for the classification of cause and effect time series generated using coupled autoregressive processes, coupled 1D chaotic skew tent maps, coupled 1D chaotic logistic maps and a real-world prey-predator system. In the case of coupled skew tent maps, the proposed method consistently outperforms a five layer Deep Neural Network (DNN) and Long Short Term Memory (LSTM) architecture for unidirectional coupling coefficient values ranging from 0.1 to 0.7. Further, we investigate the preservation of causality in the feature extracted space of NL using Granger Causality for coupled autoregressive processes and Compression-Complexity Causality for coupled chaotic systems and real-world prey-predator dataset. |
| Researcher Affiliation | Academia | Harikrishnan N. B. Department of Computer Science & Information Systems and APPCAIR BITS Pilani K. K. Birla Goa Campus, India harikrishnannb@goa.bits-pilani.ac.in Aditi Kathpalia Department of Complex Systems Institute of Computer Science of the Czech Academy of Sciences Prague, Czech Republic kathpalia@cs.cas.cz Nithin Nagaraj Consciousness Studies Programme National Institute of Advanced Studies Bengaluru, 560012, Karnataka, India nithin@nias.res.in |
| Pseudocode | No | The paper describes the methods in prose and equations but does not include explicit pseudocode blocks or sections labeled 'Algorithm'. |
| Open Source Code | Yes | The codes used in this study are available here: https://github.com/Harikrishnan NB/cause-effect-preservation-nl. |
| Open Datasets | Yes | We used simulated datasets from (a) Coupled autoregressive (AR) processes, (b) Coupled 1D chaotic maps in master-slave configuration (1D skew tent maps and 1D logistic maps) and real-world dataset from a (c) prey-predator system. The data consists of 71 data points of predator (Didinium nasutum) and prey (Paramecium aurelia) populations [35, 36]. |
| Dataset Splits | No | Table 1 specifies 'Traindata' and 'Testdata' splits but does not mention a distinct 'validation' split for hyperparameter tuning or early stopping. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., CPU, GPU models) used for running the experiments. The authors' self-assessment also states 'N/A' for compute resources. |
| Software Dependencies | No | Software implementation is performed using Python 3, scikit-learn [32], keras [33], Chaos FEX toolbox [22], Multivariate Granger Causality (MVGC) toolbox [34], CCC toolbox [8] and MATLAB. While Python 3 is mentioned, specific version numbers for scikit-learn, Keras, and the various toolboxes are not provided in the text, hindering reproducibility. |
| Experiment Setup | Yes | In the case of Chaos Net, there are three hyperparameters initial neural activity (q), discrimination threshold (b), and noise intensity (ϵ) [22]. For a fixed value of b = 0.499, and ϵ = 0.171, q was varied from 0.01 to 0.98 with a stepsize of 0.01... We choose q = 0.56 for further experiments. A five layer Deep Learning architecture was used to evaluate the efficacy of cause-effect classification. The number of nodes in the input layer = 2000, followed by first hidden layer with 5000 neurons and sigmoid activation function... Training was done for 30 epochs. |