William Murphy Email & Phone Number

City Of London, England, United Kingdom

Time Series Forecasting with Machine Learning:Deep learning models can struggle to extract information from long input sequences. The RNN for example forgets distant time steps due to it’s recursive nature and vanilla transformers are unable to process long sequences because of their O(L2) complexity. I have taken innovations from other, highly active fields, such as natural language processing and applied them to time series forecasting to create a transformer with linear complexity and which processes seasonal periods efficiently. My work outperforms state of the art models in some datasets and I hope to publish shortly.

Guildford, United Kingdom

Developed software-defined radios. Supported and extended a very large codebase. Worked as part of an agile team.

Guildford, United Kingdom

- Used deep learning techniques to recognise cancer within mammograms.- Wrote a paper titled “Using Transfer Learning for a Deep Learning Model Observer”- Paper was successfully submitted to SPIE and accepted for an oral presentation.- Experience with PyTorch, Keras, TensorFlow, Python, Linux.

Basingstoke

Over the course of my year at Sony I worked on three different solo projects:Machine learning:- Looked into the potential use of machine leaning to identify crime- Gained a strong grounding in the field of machine learning through the use of Tensorflow and python-numpy-Presented results to the departmentVirtual reality automotive show room:- Demonstrated the product at two trade shows in the Netherlands, ISE and ITEC- Used the game engine Unity to create a showroom for customers to view and customise cars- Built and optimised for both Android and PlayStation platforms- Implemented innovative features such as using VRED to create photo realistic screen shots for the userDecoding ST2110 (uncompressed video and audio over an IP network):- Used C++ to capture and process the IP packets- Boosted performance through the use of Intel intrinsic instructions (SIMD)

Gloucestershire

- Reverse engineered an electronic shower controller with the goal of developing a more advanced interface - Analysed the performance of different electronic shower mixers- Tested existing systems and discovered a security vulnerability in a casino slot machine for that had been present for 10+ years