The document summarizes research on using an extreme learning machine with local receptive fields (ELM-LRF) for face recognition. ELM-LRF was tested on three face datasets: Caltech, achieving 98.15% accuracy; CBCL, achieving 98.34% accuracy; and UFI, achieving 66.11% accuracy. ELM-LRF has advantages over other methods like reduced training time, fast results, no getting stuck in local minima, and finding the best weights in one iteration due to local connections between input and output. The research concludes that ELM-LRF is well-suited for face recognition systems.