MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202631085755 A) filed by Narula Institute Of Technology on July 13, 2026, for “adaptive Hybrid Machine Learning Framework For -Awconartee Xtmodel Selection And Optimizatio N”.
Inventor includes Ms. Debrupa Pal.
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: “Adaptive Hybrid Machine Learning Framework for Co-nAwtextare Model Selection and Optimization” ABSTRACT The present invention relates to an adaptive and intelligent machine learning framework designed to automatically select, integrate, and optimize different types of machine learning paradigms based on the intrinsic characteristics of input data, thereby overcoming the limitations of conventional static model selection approaches. The system comprises a data acquisition and preprocessing module that cleans, transforms, and standardizes heterogeneous datasets, followed by a data profiling unit that analyses key attributes such as data type, feature distribution, dimensionality, and missing values. Based on this analysis, a novel meta-learning controller dynamically determines the most suitable learning strategy, including supervised, unsupervised, semi-supervised, reinforcement learning, or a hybrid combination thereof, ensuring context-aware decision-making without human intervention. The invention further incorporates a hybrid model engine capable of integrating multiple algorithms, such as clustering techniques for pattern discovery and classification or regression models for predictive tasks, thereby enhancing overall model performance and robustness. A dynamic optimization layer is included to perform automated hyperparameter tuning, model validation, and performance benchmarking using advanced techniques such as cross-validation and iterative refinement, ensuring that the selected models operate at optimal efficiency. Additionally, the system features a continuous feedback mechanism that enables real-time learning and adaptation by incorporating new data inputs, thus improving predictive accuracy and maintaining relevance in dynamic environments. An explainability module is also integrated into the framework to provide transparent and interpretable outputs, including feature importance, decision pathways, and rationale behind model selection, which is particularly critical in high-stakes domains such as healthcare, finance, and autonomous systems. The proposed invention demonstrates significant improvements in terms of accuracy, scalability, automation, and computational efficiency compared to traditional single-model machine learning systems, as it effectively leverages the strengths of multiple learning paradigms within a unified architecture. The system is highly versatile and can be deployed across a wide range of applications, including disease prediction, fraud detection, smart IoT systems, and intelligent recommendation engines, where it can handle complex, large-scale, and evolving datasets with minimal manual intervention. By enabling seamless integration of diverse machine learning approaches and continuous self-optimization, the invention represents a transformative advancement in artificial intelligence, paving the way for the development of next-generation intelligent systems capable of autonomous learning, adaptive decision-making, and enhanced predictive intelligence in real-world scenarios. (Figure 1)
Disclaimer: Curated by HT Syndication.
