New Fractional Ambiguity Function Integrated with CNN-Based Machine Learning for Signal Classification

2026-06-06 · Source: Machine Learning · Field: Science & Research — Mathematics & Computational Sciences, Artificial Intelligence & Machine Learning · Depth: Expert, quick

Summary

A new fractional ambiguity function (NFrAF), derived from the fractional Fourier transform, is presented as a generalization of the classical ambiguity function. This NFrAF's fundamental analytical properties, including symmetry, marginality, and Moyal type identities, are rigorously established. The research verifies its capability to detect and localize both monocomponent and multicomponent linear frequency modulated (LFM) signals. Crucially, the NFrAF is integrated into a convolutional neural network (CNN)-based machine learning framework for enhanced signal classification. It provides a more informative input representation due to its superior time-frequency resolution and localization compared to conventional methods like the spectrogram and classical ambiguity function. Experimental results on simulated datasets consistently demonstrate improved classification accuracy, confirming its effectiveness for data-driven signal analysis.

Key takeaway

For Machine Learning Engineers developing signal classification systems, you should consider integrating the new fractional ambiguity function (NFrAF) as an input representation. Its demonstrated superior time-frequency resolution and localization capabilities can significantly improve classification accuracy for linear frequency modulated (LFM) signals compared to conventional spectrograms or classical ambiguity functions. Evaluate NFrAF for your CNN-based signal analysis pipelines to enhance model performance.

Key insights

The NFrAF offers superior signal representation for CNN-based classification by enhancing time-frequency resolution over traditional methods.

Principles

• Fractional ambiguity functions generalize classical methods.
• Superior time-frequency resolution improves signal classification.
• Data-driven analysis benefits from informative input representations.

Method

The NFrAF is derived from the fractional Fourier transform, its properties established, and then integrated as an input representation into a convolutional neural network for signal classification.

In practice

• Use NFrAF for LFM signal detection and localization.
• Apply NFrAF as CNN input for improved classification.
• Explore NFrAF for data-driven signal analysis tasks.

Topics

• Fractional Ambiguity Function
• Signal Classification
• Convolutional Neural Networks
• LFM Signals
• Time-Frequency Analysis
• Machine Learning

Best for: AI Scientist, Machine Learning Engineer, Research Scientist

Related on AIssential

• Neural Field Tokenizations with Hierarchy and Spatial Locality Priors — LH-NeF efficiently learns general-purpose neural field representations by integrating locality and hierarchy priors via a single forward pass.
• The Role of Ambiguity in Error Prediction via Uncertainty Quantification — Disentangling input ambiguity from UQ signals significantly improves LLM error prediction, especially for Question Answering tasks.
• Chaotic Oscillator Networks for Classification Tasks — Machine learning can approximate coupling terms in chaotic oscillator networks for scalable, efficient classification and pattern recognition.
• The Score Kalman Filter — The Score Kalman Filter enables scalable nonlinear Bayesian filtering by replacing intractable partition function evaluations with efficient linear algebra.
• Self-Supervised Learning via Flow-Guided Neural Operator on Time-Series Data — FGNO enhances self-supervised learning for time-series data by dynamically varying corruption levels and using clean inputs for inference.
• Certified and accurate computation of function space norms of deep neural networks — Certified function space norms for neural networks are achieved via interval arithmetic and adaptive quadrature.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning.