This paper presents an all-optical spiking neurosynaptic network capable of supervised and unsupervised learning. The system consists of an all-optical spiking neuron connected to photonic synapses via an integrated photonics network, enabling a small-scale neurosynaptic system for pattern recognition. The network uses wavelength division multiplexing (WDM) to implement a scalable photonic neural network with 140 optical elements, demonstrating pattern recognition in the optical domain. The system is based on phase-change materials (PCMs) that provide the integrate-and-fire functionality of neurons and synaptic plasticity. The neuron circuit is designed to mimic biological neurons, with synaptic weights adjusted through feedback loops for both supervised and unsupervised learning. Supervised learning uses error backpropagation, while unsupervised learning employs spike-timing-dependent plasticity (STDP). The system is implemented on a nanophotonic chip, enabling high-speed, low-energy processing. The network is capable of recognizing patterns, such as letters, and can be scaled to more complex tasks, including language identification. The all-optical approach offers high bandwidth and fast signaling, making it suitable for processing telecommunication and visual data. The system is built using electron-beam lithography and phase-change materials, with photonic circuits designed for efficient signal processing. The network is scalable and can be integrated with off-chip components, offering potential for future neuromorphic systems. The work demonstrates the feasibility of all-optical neurosynaptic networks for efficient, low-energy computation.This paper presents an all-optical spiking neurosynaptic network capable of supervised and unsupervised learning. The system consists of an all-optical spiking neuron connected to photonic synapses via an integrated photonics network, enabling a small-scale neurosynaptic system for pattern recognition. The network uses wavelength division multiplexing (WDM) to implement a scalable photonic neural network with 140 optical elements, demonstrating pattern recognition in the optical domain. The system is based on phase-change materials (PCMs) that provide the integrate-and-fire functionality of neurons and synaptic plasticity. The neuron circuit is designed to mimic biological neurons, with synaptic weights adjusted through feedback loops for both supervised and unsupervised learning. Supervised learning uses error backpropagation, while unsupervised learning employs spike-timing-dependent plasticity (STDP). The system is implemented on a nanophotonic chip, enabling high-speed, low-energy processing. The network is capable of recognizing patterns, such as letters, and can be scaled to more complex tasks, including language identification. The all-optical approach offers high bandwidth and fast signaling, making it suitable for processing telecommunication and visual data. The system is built using electron-beam lithography and phase-change materials, with photonic circuits designed for efficient signal processing. The network is scalable and can be integrated with off-chip components, offering potential for future neuromorphic systems. The work demonstrates the feasibility of all-optical neurosynaptic networks for efficient, low-energy computation.