Innovative Projects Realized

Toward an Understanding of Beautiful Feather Cover in Laying Hens – Year Two

Feather pecking (FP) in egg-laying hens, where individuals peck at other birds to pull out and eat their feathers, is a challenge for the sector with large economic and welfare implications. It is especially of concern in systems where birds are housed in large social groups as it is harder to control.
With new policies in Canada leading to the transition from conventional cage to alternative housing systems, it becomes imperative to reduce the risk of large scale FP outbreaks. This project aims to develop a Canadian FP Management Plan (CFMP) by identifying risk factors for FP in alternative housing systems while developing an illustrated guide for farmers to assess plumage condition. This will be translated into the CFMP and provide the Egg Farmers of Canada (EFC) with advice on how to prevent/redu

Improving Metallic Yield in a Steel Rolling Plant through Optimization

The objective of this project is to use optimization to improve metallic yield (the percentage of raw material that ends up as usable product) in an ArcelorMittal Steel Rolling Plant. The metallic yield of the rolling operations depends upon the length of billets from which the final product is manufactured. Ideally, a single customer order would be filled using billets of precisely the length that would yield the minimum achievable amount of scrap. However, ideal yields for each order in a set of customer

orders cannot be aggregated to fulfill them at the ideal yield for the entire set of orders, as this conflicts with the objective of keeping the inventory at a minimum level. In other words, the optimal way of fulfilling a set of customer orders is a tradeoff between yield maximization and inventory minimization. This project will develop efficient and robust models and algorithms for optimal billet ordering…

Toward an Understanding of Beautiful Feather Cover in Laying Hens

Feather pecking (FP) in egg-laying hens, where individuals peck repetitively and excessively at other birds to pull out and eat their feathers, is a challenge for the industry with large economic and welfare implications. High prevalence of FP is reported (60-80%) and this is associated with mortality rates of up to 20-40%, which translates to hundreds of millions of birds dying due to FP every year. It is especially of concern in systems where birds are housed together in large social groups as it is harder to control.

With new policies in Canada leading to the transition from conventional cage to alternative housing systems, it becomes imperative to reduce the risk of large scale FP outbreaks. This proposal aims to develop a Canadian FP Management Plan (CFMP) to ensure safe and successful transition to alternative systems. Therefore, we will identify Canadian-tailored risk factors for FP in alternative housing systems through questionnaires while developing an illustrated guide for farmers/auditors to assess plumage condition. This knowledge will be translated into the CFMP and this tool will provide advice on courses of action to prevent/reduce/stop FP in Canadian hen flocks allowing for transitioning to alternative housing systems while maintaining high animal welfare standards.

Applications of deep learning to large-scale data analysis in mass spectrometry-based proteomics – Year Two

As a result of recent advances in high-throughput technologies, rapidly increasing amounts of mass spectrometry (MS) data pose new opportunities as well as challenges to existing analysis methods. Novel computational approaches are needed to take advantage of latest breakthroughs in high-performance computing for the large-scale analysis of big data from MS-based proteomics. In this project, we aim to develop new applications of deep learning and neural networks for the analysis of MS data. In particular, we focus on three fundamental problems in a typical MS analysis workflow: peptide feature detection and quantification, de novo peptide sequencing, and protein identification and quantification. Once successfully evaluated, the proposed techniques will be implemented and integrated to PEAKS Studio, the current MS analysis platform of the partner. We believe that the project results will contribute major advances to the research field of MS-based proteomics and substantially improve the performance of the partner’s software products.

Applications of deep learning to large-scale data analysis in mass spectrometry-based proteomics

Rapidly increasing amounts of mass spectrometry (MS) data pose new opportunities as well as challenges to existing analysis methods. Novel computational approaches are needed to take advantage of latest breakthroughs in high-performance computing for the large-scale analysis of big data from MS-based proteomics. In this project, we aim to develop new applications of deep learning and neural networks for the analysis of MS data. In particular, we focus on three fundamental problems in a typical MS analysis workflow: peptide feature detection and quantification, de novo peptide sequencing, and protein identification and quantification. Once successfully evaluated, the proposed techniques will be implemented and integrated to PEAKS Studio, the current MS analysis platform of the partner.

Using Deep Learning to Auto-tune GPU Application

The fellowship mainly investigates an analysis of the state-of-the-art approaches, design and implementation of cutting-edge deep neural network models to be used on a mobile platform. It explored ways to optimize the deployment of these machine-learning models for prediction tasks on the mobile devices which requires energy efficiency and accuracy.

Using Deep Learning for Auto-tuning of High Performance GPU Applications

Graphics Processing Units (GPUs) are increasingly used to accelerate applications and to reduce their energy use. GPUs are particularly attractive for mobile platforms, where battery life is important. However, GPUs are hard to use, requiring developers to apply optimizations to their code to realize the performance and energy benefits of GPUs–a tedious and error prone process.

This project involves the analysis, implementation and evaluation of a state-of-the-art machine learning based framework to automatically determine what optimizations to apply to GPU programs. Specifically, we plan to explore cutting-edge deep learning methods that have shown great success in recent years in domains such as image processing, computer vision and text processing. Yet, there is little work applying them to performance auto-tuning.

Adapting learning methods to our tuning problem is a hard task, requiring several challenges to be tackled: availability of training data, determining optimal model parameters and dealing with computational complexity. We will build upon our earlier successes in the field to tackle these challenges. The results of this project will benefit our industrial partner, Qualcomm, a key player in the GPU industry. It will also address important problems of optimizations for GPU applications that are outstanding in the community.

Computational and experimental characterization of mechanical performance of cross laminated timber (CLT)

Cross-laminated timber (CLT) is an engineered wood panel typically consisting of multiple layers of glued timber stacked in a cross-ply layup. Timber shows a strong anisotropic mechanical behavior due to its microstructure. With a cross lamination, the CLT possesses superior dimensional stability, strength and rigidity, in comparison to traditional wood products. In Canada, CLT is gaining increasing recognition as a high-performance material for structural systems, as well as a new opportunity for wood in non-traditional applications. In order to fulfill its potential applications, the mechanical performance of the CLT needs to be studied in details, especially the failure mechanisms. The objective of this project is to systematically investigate and characterize the mechanical performance of the CLT through both numerical modelling and experimental methods. Successful completion of this project will provide innovative solutions for the industry partner, Guardian Structures, to help them achieve their desired mechanical performance of CLT and pave the path to new applications.

Development of an information theory-based mutation detector for a commercial bioinformatics genome server

I have recently developed a piece of software that can be used to interpret the effects of DNA sequence differences in human genomes. The analysis produces results that predict disease mutations. Dr. Rogan’s laboratory has developed approaches of visualizing DNA sequence data, which I will incorporate into this software. I will modify the existing visualization software to run Java and integrate this Java code in to my previously developed mutation detection software. This proposal will also improve the performance of the software I developed. This is important because analysis of large numbers of mutations currently takes many hours on a typical computer server, and the user expects these results more quickly. The software currently analyzes variants that alter gene expression. The capabilities of this software will also be extended to analyze others types of mutations. This software will allow researchers to focus their efforts on confirming only genetic variants….

Distant Pointing in Virtual Environments

No Project Overview was submitted for this so here is me rambling on to finnish the wordcount.

Improving Powder Performance by Development and Optimization of Industrial Lubricants and Mixing Technology for Powder Metallurgy – Year two

Ideal flow, high-volume Powder Metallurgy (PM) manufacturing can achieve uniform, consistent filling of die cavities, leading to high productivity, low rejection rates, improved part integrity and consistent part dimensions. The type and amount of lubricant, size and shape of lubricant particles, mixing parameters and certain environmental conditions all significantly influence the flow characteristics and apparent density (AD) of powder mixtures. Lubricants also affect part integrity and strength after compaction, and must be chosen carefully to ensure high part material density after pressing. Properly controlling delubrication and sintering conditions helps avoid stain formation and improves the mechanical properties of final products.
Industrial lubricants produced by H.L. Blachford are mixed with iron powder produced by Rio-Tinto to make “press-ready” powder mixtures for PM parts. TO BE CONT’D

Improving Powder Performance by Development and Optimization of Industrial Lubricants and Mixing Technology for Powder Metallurgy

Ideal flow, high-volume Powder Metallurgy (PM) manufacturing can achieve uniform, consistent filling of die cavities, leading to high productivity, low rejection rates, improved part integrity and consistent part dimensions. The type and amount of lubricant, size and shape of lubricant particles, mixing parameters and certain environmental conditions all significantly influence the flow characteristics and apparent density (AD) of powder mixtures. Lubricants also affect part integrity and strength after compaction, and must be chosen carefully to ensure high part material density after pressing. Properly controlling delubrication and sintering conditions helps avoid stain formation and improves the mechanical properties of final products.
Industrial lubricants produced by H.L. Blachford are mixed with iron powder produced by Rio-Tinto to make “press-ready” powder mixtures for PM parts. TO BE CONT’D