My final year

Hello everyone,

It has been a while since I last wrote. This is because I am nearing the end of my PhD and to be honest I found myself quite busy with writing papers to publish and wrapping up experiments for my project and hence it has been a ghost town here since my last post. I will be submitting my thesis this September this year.  As they say, all good things must come to an end, just like Episode 3 – Season 8 of Game of Thrones 🙂

Four years of discovering and learning new things and trying to push the boundaries in my field have all culminated to this very moment where I have to wrap these years into a 150 page or so thesis. On reflection, I feel that I have grown a lot as a person and a researcher during my time here. I started my journey at Manchester during my Masters where I came to learn more about Artificial Intelligence. During my master’s program, I was exposed to topics in machine learning, computer vision, text mining, but unexpectedly, I came across this module on parallel programming that drew my curiosity even more. This module helped me understand that current computing systems such as those found in our laptops and smartphones use multiple processors and how such systems can be programmed.

After I completed my masters program, I thus found a deep interest in understanding the field of Artificial Intelligence from a computer architecture’s perspective. This led me to join The Advanced Processor’s Technology Group and to choose a project that sits at the junction of machine learning and computer architecture research. My first task during my PhD, was surveying the literature to see how existing machine learning, in particular deep learning algorithms are implemented on real systems. Interestingly, most of deep learning research was actually being done on the cloud and people were looking at making deep learning more efficient to run locally on devices, for example, your smartphone. Google’s on-device Speech Recognizer, is a recent example that shows the potential of this field.

Now, this field has grown tremendously in the past four years, where, I am seeing a lot more people getting involved in bringing machine learning to run on  a device. I am now organising a workshop called  Green Data Mining that brings researchers together and foster collaborations among people working on energy-efficient machine learning.

As a researcher, I can now say that through my PhD work, I have dipped my toes in this field by contributing to tools SyNERGY that can measure the energy of deep learning algorithms on current mobile devices. I have used this tool to build predictive models so that researchers can now predict the energy consumption on a particular device without needing to measure it.

Apart from this, there is a lot of buzz around building the next generation processor for deep learning. I was exposed to this area during my internship at ARM where I carried forward this work into my PhD. Here is a link to my recently published paper.

Finally, I do feel, however, that I should give a snapshot view interesting events that I enjoyed outside my own bubble of research.

I think one of the most amazing moments was when the Department of Computer Science celebrated the count down towards the one million-th processor Neuromorphic Supercomputer! I am quite proud of all of my friends and colleagues from the Advanced Processor Technology research group who were part of this project!

Around the same time, a new research group devoted to Cognitive Robotics became part of the University of Manchester. Now, I have befriended some friendly humans and robots!

I am helping out with an experiment related to intention reading with a humanoid robot called iCub

The video for the experiment can be found here.

The Department of Computer Science also saw its first band, named “The Kill and Burn ”. The band’s name comes from, well, Tom Kilburn after whom the computer science building is named.

The Kill and Burn performing the common room, Kilburn Building

It was good to see them perform at the common room for Christmas and brought some much needed cheer before the holidays.

Finally, I took part in the yearly MUBLS dance society dance competitions for the first time that truly was inspiring to watch and fun to participate in.

Leicester Competition 2018
Manchester Winter Competition 2018

 

I will see you next time with a final update on my next adventure.

Cheers,

Crefeda