Kavishka Abeywardhana’s Post

Machine Learning Based Featureless Signalling: A Brief Review 📡 In the field of secure communication, challenges like jamming and interference are common. Traditional methods such as Direct-sequence spread-spectrum (DSSS) have been used to improve security by broadening signal bandwidth. However, vulnerabilities exist, as attackers can exploit the periodicity and autocorrelation of pseudo-random noises. 📶 A machine learning approach can revolutionize spectrum spreading by treating the communication system as an autoencoder. Through optimization, signals are transformed to resemble Gaussian samples, blending into background noise for heightened security. Explore this innovative concept in greater detail in the article sourced from, Shakeel, I. (2018), titled "Machine Learning Based Featureless Signalling" available on ArXiv: /abs/1807.07260.

Discrete Fourier Series (DFS) is a crucial technique in digital signal processing, deriving Fourier representations for discrete-time periodic sequences. This blog post explains the fundamentals of DFS based on the renowned work "Discrete-Time Signal Processing" by Alan V. Oppenheim, Ronald W. Schafer, and John R. Buck. #DigitalSignalProcessing #FourierTransform #DFS #SignalProcessing #TechBlog

An Introduction to Deep Learning for the Physical Layer: A Brief Review Exploring deep learning and machine learning in enhancing modern communication systems is crucial. The shift from traditional expert knowledge-based systems to more advanced models like autoencoders for the transmitter, channel, and receiver combination is pivotal. In my latest blog post, I discussed this concept along with innovations such as radio transmitter networks. This article is inspired by the "Introduction to Deep Learning for the Physical Layer" paper authored by Timothy O’Shea and Jakob Hoydis. J., T., & Hoydis, J. (2017). An Introduction to Deep Learning for the Physical Layer. ArXiv. /abs/1702.00832 #DeepLearning #MachineLearning #CommunicationSystems #RadioTransmitterNetworks #Innovation

Bayesian Statistics: A Foundational Approach for Machine Learning Bayesian statistics plays a critical role in machine learning and data science. Its applications span representation learning, unsupervised learning, and probabilistic models. This core concept involves updating our beliefs about the world based on new evidence. It offers a compelling alternative to traditional frequentist thinking. In this post, I have provided an intuitive introduction to Bayesian statistics, including detailed explanations of maximum a posteriori probability (MAP), mean squared error (MSE), linear approximations, and Bayes' theorem.

Probability Theory for Machine Learning 🎲 Understanding probability is crucial in the realm of machine learning. It forms the foundation, aiding in understanding randomness and quantifying uncertainty to enhance ML models effectively. Excited to share a comprehensive tutorial explaining the fundamentals of machine learning. Covering essential topics like random variables, independence, conditional probability, probability distributions, special functions, and the Bayes rule. #MachineLearning #ProbabilityTheory #DataScience #AI #TechEducation

Linear Factor Models (LFM) 📊 LFMs are among the earliest probabilistic models designed to generate samples from a data distribution. They employ latent variables to encapsulate the underlying structure of the dataset. In this blog post, I have explained the fundamentals of probabilistic PCA, Independent Component Analysis (ICA), Sparse coding, and Slow feature analysis. #DataScience #LinearFactorModels #ProbabilisticModels