PhD Candidate in Computer Architecture with Specific Focus on Quantized Neural Networks

Ultra-fast Recurrent Neural Networks with SambaNova's Reconfigurable Dataflow Architecture

We have a vacancy for a PhD candidate within the area of computer architecture at the Department of Computer Science.

For a position as a PhD Candidate, the goal is a completed doctoral education up to an obtained doctoral degree.

Artificial neural networks have emerged as a solution to a wide variety of problems, such as medical diagnosis, object tracking, voice recognition, translation etc. A drawback with artificial neural networks is that they tend to be computationally heavy resulting in the need for specialized hardware that requires resources (silicon area) and energy. Quantized neural networks (QNN) is one solution for reducing the computational demand. By quantizing the weights and activations (i.e., using lower precision) the accelerator can be made smaller, and less memory is needed for storing weights and activations, resulting in lower energy usage.

The objective of this PhD project is to investigate quantized neural networks, their training, and implementation as accelerators for energy constrained devices, e.g., devices at the edge and Internet of Things (IoT). Our starting point will be our existing work on bit-serial accelerators, such as BISMO, and we will adapt frameworks for training low-precision QNNs, like the Brevitas framework from Xilinx. The goal is to create complete and efficient solutions for constrained devices.

Your immediate leader is the Head of the Computing unit

Duties of the position

Conducting high quality research and report progress on a regular basis in agreement with the supervisors.
Undertake the necessary courses (30 ECTS credits) as part of the PhD program.
Submit an application for admission and a research plan no later than 3 months after employment.
Perform duty work at the department, e.g., as a teaching assistant, at 25% of the time.

Required selection criteria

You must have a master’s degree in Computer Science, Electrical Engineering, or equivalent. Students expecting to graduate with a relevant master’s degree by the end of 2022 are encouraged to apply.

Your education must correspond to a five-year Norwegian degree programme, where 120 credits are obtained at master’s level.

You must have a strong academic background from your previous studies and an average grade from the master’s degree program, or equivalent education, which is equal to B or better compared with NTNU’s grading scale. If you do not have letter grades from previous studies, you must have an equally good academic basis. If you have a weaker grade background, you may be assessed if you can document that you are particularly suitable for a PhD education.

You must meet the requirements for admission to the faculty’s doctoral program.
You must have strong programming skills.
You must have good written and oral English skills.
The appointment is to be made in accordance with Regulations concerning the degrees of Philosophiae Doctor (PhD) and Philosodophiae Doctor (PhD) in artistic research national guidelines for appointment as PhD, post doctor and research assistant.

Preferred selection criteria

Strong background in computer architecture.
Experience with artificial neural networks and their training.
Experience with hardware design at the Register Transfer Level (RTL), preferably with Chisel.
Experience with using reconfigurable computing platforms such as Field Programmable Gate Arrays (FP
GAs).
Experience with compiler infrastructures.

Personal Characteristics

Communicates information with clarity and ease, both orally and in writing.
Learns from experience and feedback from others.
Sees the big picture and takes broader considerations into account.
Acquires new knowledge quickly and can use existing knowledge in new ways.
Ability to work independently but also in a team.

About the application

The application and supporting documentation to be used as the basis for the assessment must be in English.

Publications and other scientific work must follow the application. Please note that your application will be considered based solely on information submitted by the application deadline. You must therefore ensure that your application clearly demonstrates how your skills and experience fulfil the criteria specified above.

The application must include:

A cover letter where you express your particular interest and suitability for the position (maximum 1 page).
CV, certificates, and diplomas.
Transcripts and diplomas for bachelor’s and master’s degrees. If you have not completed the master’s degree, you must submit the transcripts of all courses passed so far.
A copy of the master’s thesis. If you recently have submitted your master’s thesis, you can attach a draft of the thesis. Documentation of a completed master’s degree must be presented before taking up the position.
Name and address of three referees.
Publications or other relevant research work (if any).

If all, or parts, of your education has been taken abroad, we also ask you to attach documentation of the scope and quality of your entire education, both bachelor’s and master’s education, in addition to other higher education. Description of the documentation required can be found here. If you already have a statement from NOKUT, please attach this as well.

We will take joint work into account. If it is difficult to identify your efforts in the joint work, you must enclose a short description of your participation.

In the evaluation of which candidate is best qualified, emphasis will be placed on education, experience and personal and interpersonal qualities. Motivation, ambitions, and potential will also count in the assessment of the candidates.

NTNU is committed to following evaluation criteria for research quality according to The San Francisco Declaration on Research Assessment – DORA.

APPLY NOW