Ijraset Journal For Research in Applied Science and Engineering Technology
Authors: Sudip Barua
DOI Link: https://doi.org/10.22214/ijraset.2025.72951
Certificate: View Certificate
In this work, we propose a novel boosting-based machine learning algorithm called EvoBoost, invented by Sudip Barua. Gradient boosting has emerged as a cornerstone technique in machine learning, achieving state-of-the-art performance in both classification and regression tasks. While existing models such as XGBoost, LightGBM, and CatBoost are widely adopted, they present challenges including excessive hyperparameter tuning, high memory consumption, and suboptimal handling of imbalanced data. EvoBoost addresses these limitations through a streamlined boosting framework that is both effective and easy to implement. It introduces probabilistic residuals for classification and a clean, interpretable residual computation for regression. Extensive empirical evaluations across six benchmark datasets demonstrate that EvoBoost consistently outperforms or matches the performance of established models in terms of accuracy, R² score, and log loss, while maintaining superior interpretability and implementation simplicity.
Gradient boosting models like XGBoost, LightGBM, and CatBoost have become mainstream for classification and regression due to their high accuracy. However, they suffer from:
Complex hyperparameter tuning
High memory consumption
Poor handling of imbalanced datasets
Reduced interpretability
To overcome these limitations, Sudip Barua introduced EvoBoost, a new gradient boosting algorithm that is simpler, interpretable, and efficient, while delivering competitive performance.
Uses decision tree regressors in an iterative boosting loop.
Employs residual learning to correct errors at each stage.
For regression, residuals = actual − predicted.
For classification, residuals = true label (one-hot) − predicted probabilities (via softmax).
Avoids complex operations like second-order gradients or feature binning.
Reduces the need for extensive tuning and is more accessible for domain experts.
The EvoBoost algorithm follows a 5-step iterative process:
Initialize the model with a baseline prediction.
Compute residuals from current predictions.
Train a tree on these residuals.
Update the model with the new tree's output (scaled by learning rate).
Stop training when performance on a validation set stops improving.
XGBoost: Accurate but tuning-heavy; sensitive to class imbalance.
LightGBM: Fast and memory-efficient but may overfit small or noisy datasets.
CatBoost: Good for categorical data but complex and resource-heavy.
EvoBoost++: Aims for a middle ground—simple, robust, and understandable.
Boosting is viewed as gradient descent in function space.
Regression minimizes Mean Squared Error.
Classification minimizes Cross-Entropy Loss.
Trees approximate the negative gradient at each step, pushing predictions toward true values or class probabilities.
The approach ensures fast convergence and broad applicability.
EvoBoost was tested on six datasets (three regression and three classification).
Benchmarked against leading models.
Demonstrated competitive accuracy, with the added benefits of interpretability, stability, and lower complexity.
In this study, we introduced EvoBoost, a unified gradient boosting algorithm invented by Sudip Barua that prioritizes simplicity, interpretability, and high performance across both regression and classification domains. Unlike traditional gradient boosting implementations that rely on complex heuristics and second-order approximations, EvoBoost adopts a principled first-order gradient descent framework based on intuitive residual learning. By integrating probabilistic softmax residuals for classification and direct error minimization for regression, EvoBoost provides a cohesive approach that adapts well to a variety of datasets. Through extensive experimentation across six benchmark datasets—three classification and three regression—we demonstrated that EvoBoost consistently delivers competitive or superior results compared with well-established models such as XGBoost, LightGBM, and CatBoost. Notably, EvoBoost excels in producing robust predictions on imbalanced and noisy datasets, highlighting its generalization capacity. It achieves this while maintaining a user-friendly structure that avoids the pitfalls of excessive hyperparameter tuning, specialized encoders, or GPU-only execution paths. Moreover, EvoBoost has been constructed to meet the growing demand for explainable AI. Its use of decision tree regressors allows users to visualize splits, assess feature importance, and interpret outcomes with minimal effort—key requirements in sensitive domains such as healthcare, finance, and legal analytics. This interpretability is complemented by its minimal memory overhead and fast training times, which make it suitable for edge devices, real-time inference, and academic settings. We believe EvoBoost, invented by Sudip Barua, is well-positioned to inspire future research into interpretable ensemble learning. Future work will involve formalizing its uncertainty estimation capabilities, extending it to unsupervised and semi-supervised learning, and implementing GPU-accelerated variants for large-scale industrial use. Additionally, plans are underway to release an open-source library that facilitates rapid experimentation and seamless integration into existing ML pipelines. In conclusion, EvoBoost exemplifies the next step in gradient boosting evolution—one that harmonizes power with clarity, and performance with accessibility. We invite researchers and practitioners to adopt, critique, and enhance EvoBoost, paving the way toward more trustworthy and scalable AI systems.
[1] Friedman, J. H. (2001). Greedy function approximation: A gradient boosting machine. Annals of Statistics, 29(5), 1189-1232. [2] Chen, T., & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785-794. [3] Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W., ... & Liu, T. Y. (2017). LightGBM: A highly efficient gradient boosting decision tree. Advances in Neural Information Processing Systems, 30, 3146-3154. [4] Dorogush, A. V., Ershov, V., & Gulin, A. (2018). CatBoost: gradient boosting with categorical features support. arXiv preprint arXiv:1810.11363. [5] Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., ... & Duchesnay, É. (2011). Scikit-learn: Machine learning in Python. Journal of Machine Learning Research, 12, 2825-2830. [6] Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5-32. [7] Hastie, T., Tibshirani, R., & Friedman, J. (2009). The Elements of Statistical Learning (2nd ed.). Springer. [8] Freund, Y., & Schapire, R. E. (1997). A decision-theoretic generalization of on-line learning and an application to boosting. Journal of Computer and System Sciences, 55(1), 119-139. [9] Mason, L., Baxter, J., Bartlett, P., & Frean, M. (1999). Boosting algorithms as gradient descent. Advances in Neural Information Processing Systems, 12. [10] Friedman, J. H. (2002). Stochastic gradient boosting. Computational Statistics & Data Analysis, 38(4), 367-378. [11] Natekin, A., & Knoll, A. (2013). Gradient boosting machines, a tutorial. Frontiers in Neurorobotics, 7, 21. [12] Prokhorenkova, L., Gusev, G., Vorobev, A., Dorogush, A. V., & Gulin, A. (2018). CatBoost: unbiased boosting with categorical features. Advances in Neural Information Processing Systems, 31. [13] Tyree, S., Weinberger, K. Q., Agrawal, K., & Paykin, J. (2011). Parallel boosted regression trees for web search ranking. Proceedings of the 20th International Conference on World Wide Web, 387-396. [14] Zhou, Z. H. (2012). Ensemble Methods: Foundations and Algorithms. Chapman & Hall/CRC. [15] Bühlmann, P., & Hothorn, T. (2007). Boosting algorithms: Regularization, prediction and model fitting. Statistical Science, 22(4), 477-505. [16] Chen, T., He, T., Benesty, M., Khotilovich, V., Tang, Y., Cho, H., & Chen, K. (2015). XGBoost: extreme gradient boosting. R Package Vignette. [17] Raschka, S., & Mirjalili, V. (2019). Python Machine Learning (3rd ed.). Packt Publishing. [18] James, G., Witten, D., Hastie, T., & Tibshirani, R. (2013). An Introduction to Statistical Learning. Springer. [19] Murphy, K. P. (2012). Machine Learning: A Probabilistic Perspective. MIT Press. [20] Zhang, T., & Johnson, R. (2014). Learning nonlinear functions using regularized greedy forest. IEEE Transactions on Pattern Analysis and Machine Intelligence, 36(5), 942-954. [21] Quinlan, J. R. (1986). Induction of decision trees. Machine Learning, 1(1), 81-106. [22] Ho, T. K. (1995). Random decision forests. Proceedings of 3rd International Conference on Document Analysis and Recognition, 1, 278-282. [23] Cortes, C., & Vapnik, V. (1995). Support-vector networks. Machine Learning, 20(3), 273-297. [24] Bishop, C. M. (2006). Pattern Recognition and Machine Learning. Springer. [25] Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press. [26] Bottou, L. (2010). Large-scale machine learning with stochastic gradient descent. Proceedings of COMPSTAT\'2010, 177-186. [27] Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society: Series B, 58(1), 267-288. [28] Zou, H., & Hastie, T. (2005). Regularization and variable selection via the elastic net. Journal of the Royal Statistical Society: Series B, 67(2), 301-320. [29] Breiman, L., Friedman, J. H., Olshen, R. A., & Stone, C. J. (1984). Classification and Regression Trees. Wadsworth. [30] Schapire, R. E. (1990). The strength of weak learnability. Machine Learning, 5(2), 197-227. [31] Drucker, H. (1997). Improving regressors using boosting techniques. Proceedings of the Fourteenth International Conference on Machine Learning, 107-115. [32] Ridgeway, G. (1999). The state of boosting. Computing Science and Statistics, 31, 172-181. [33] Dietterich, T. G. (2000). Ensemble methods in machine learning. International Workshop on Multiple Classifier Systems, 1-15. [34] Caruana, R., & Niculescu-Mizil, A. (2006). An empirical comparison of supervised learning algorithms. Proceedings of the 23rd International Conference on Machine Learning, 161-168. [35] Rokach, L., & Maimon, O. (2005). Top-down induction of decision trees classifiers-a survey. IEEE Transactions on Systems, Man, and Cybernetics, Part C, 35(4), 476-487. [1] Friedman, J. H. (2001). Greedy function approximation: A gradient boosting machine. Annals of Statistics, 29(5), 1189-1232. [2] Chen, T., & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785-794. [3] Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W., ... & Liu, T. Y. (2017). LightGBM: A highly efficient gradient boosting decision tree. Advances in Neural Information Processing Systems, 30, 3146-3154. [4] Dorogush, A. V., Ershov, V., & Gulin, A. (2018). CatBoost: gradient boosting with categorical features support. arXiv preprint arXiv:1810.11363. [5] Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., ... & Duchesnay, É. (2011). Scikit-learn: Machine learning in Python. Journal of Machine Learning Research, 12, 2825-2830. [6] Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5-32. [7] Hastie, T., Tibshirani, R., & Friedman, J. (2009). The Elements of Statistical Learning (2nd ed.). Springer. [8] Freund, Y., & Schapire, R. E. (1997). A decision-theoretic generalization of on-line learning and an application to boosting. Journal of Computer and System Sciences, 55(1), 119-139. [9] Mason, L., Baxter, J., Bartlett, P., & Frean, M. (1999). Boosting algorithms as gradient descent. Advances in Neural Information Processing Systems, 12. [10] Friedman, J. H. (2002). Stochastic gradient boosting. Computational Statistics & Data Analysis, 38(4), 367-378. [11] Natekin, A., & Knoll, A. (2013). Gradient boosting machines, a tutorial. Frontiers in Neurorobotics, 7, 21. [12] Prokhorenkova, L., Gusev, G., Vorobev, A., Dorogush, A. V., & Gulin, A. (2018). CatBoost: unbiased boosting with categorical features. Advances in Neural Information Processing Systems, 31. [13] Tyree, S., Weinberger, K. Q., Agrawal, K., & Paykin, J. (2011). Parallel boosted regression trees for web search ranking. Proceedings of the 20th International Conference on World Wide Web, 387-396. [14] Zhou, Z. H. (2012). Ensemble Methods: Foundations and Algorithms. Chapman & Hall/CRC. [15] Bühlmann, P., & Hothorn, T. (2007). Boosting algorithms: Regularization, prediction and model fitting. Statistical Science, 22(4), 477-505. [16] Chen, T., He, T., Benesty, M., Khotilovich, V., Tang, Y., Cho, H., & Chen, K. (2015). XGBoost: extreme gradient boosting. R Package Vignette. [17] Raschka, S., & Mirjalili, V. (2019). Python Machine Learning (3rd ed.). Packt Publishing. [18] James, G., Witten, D., Hastie, T., & Tibshirani, R. (2013). An Introduction to Statistical Learning. Springer. [19] Murphy, K. P. (2012). Machine Learning: A Probabilistic Perspective. MIT Press. [20] Zhang, T., & Johnson, R. (2014). Learning nonlinear functions using regularized greedy forest. IEEE Transactions on Pattern Analysis and Machine Intelligence, 36(5), 942-954. [21] Quinlan, J. R. (1986). Induction of decision trees. Machine Learning, 1(1), 81-106. [22] Ho, T. K. (1995). Random decision forests. Proceedings of 3rd International Conference on Document Analysis and Recognition, 1, 278-282. [23] Cortes, C., & Vapnik, V. (1995). Support-vector networks. Machine Learning, 20(3), 273-297. [24] Bishop, C. M. (2006). Pattern Recognition and Machine Learning. Springer. [25] Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press. [26] Bottou, L. (2010). Large-scale machine learning with stochastic gradient descent. Proceedings of COMPSTAT\'2010, 177-186. [27] Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society: Series B, 58(1), 267-288. [28] Zou, H., & Hastie, T. (2005). Regularization and variable selection via the elastic net. Journal of the Royal Statistical Society: Series B, 67(2), 301-320. [29] Breiman, L., Friedman, J. H., Olshen, R. A., & Stone, C. J. (1984). Classification and Regression Trees. Wadsworth. [30] Schapire, R. E. (1990). The strength of weak learnability. Machine Learning, 5(2), 197-227. [31] Drucker, H. (1997). Improving regressors using boosting techniques. Proceedings of the Fourteenth International Conference on Machine Learning, 107-115. [32] Ridgeway, G. (1999). The state of boosting. Computing Science and Statistics, 31, 172-181. [33] Dietterich, T. G. (2000). Ensemble methods in machine learning. International Workshop on Multiple Classifier Systems, 1-15. [34] Caruana, R., & Niculescu-Mizil, A. (2006). An empirical comparison of supervised learning algorithms. Proceedings of the 23rd International Conference on Machine Learning, 161-168. [35] Rokach, L., & Maimon, O. (2005). Top-down induction of decision trees classifiers-a survey. IEEE Transactions on Systems, Man, and Cybernetics, Part C, 35(4), 476-487.
Copyright © 2025 Sudip Barua. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Paper Id : IJRASET72951
Publish Date : 2025-07-01
ISSN : 2321-9653
Publisher Name : IJRASET
DOI Link : Click Here