Intelligent Architectures for HPC Platforms

Andrew McGee1

1Senior Technologist Huawei Technologies

 

 

Huawei has successfully deployed high-performance computing (HPC) clusters for its customers in industries such as manufacturing, supercomputing, and higher education through a complete series of HPC products and solutions. Huawei has claimed a position on the top rung of the global supercomputing markets and accumulated rich experience and industry practices in the AI domain, which is represented by public cloud and video analytics applications. These advantages are helping Huawei build industry-leading HPC cloud and AI convergence solutions.  Come and here more at this valuable session.

Fast and Accurate Training of an AI Radiologist on the Zenith Supercomputer

Andrew Underwood1

1Director, HPC & AI Dell EMC

 

The health care industry is expected to be an early adopter of AI and deep learning to improve patient outcomes, reduce costs, and speed up diagnosis. We have developed models for using AI to diagnose pneumonia, emphysema, and other thoracic pathologies from chest x-rays. Using the Stanford University CheXNet model as inspiration, we explore ways of developing accurate models for this problem with fast parallel training on Zenith, the Intel Xeon-based supercomputer at Dell EMC’s HPC and AI Innovation Lab. We explore various network topologies to gain insight into what types of neural networks scale well in parallel and improve training time from days to hours. We then explore transferring this learned knowledge to other radiology subdomains, such as mammography, and whether this leads to better models than developing subdomain models independently.

Building AI tools and frameworks on Dell EMC HPC Systems

Matt Wallis1

1HPC Solution Architect, Dell EMC

 

This session is a technical demonstration of deploying and managing deep learning frameworks, tools and libraries used on Dell EMC Supercomputers, with the objective of accelerating data science and artificial intelligence research projects.

The Future Impact of Sequencing Everyone

Michael McManus

 

As the use of genome sequencing becomes mainstream in clinical practice we are faced with the daunting realization of a future where we are “sequencing everyone”. What would it take to sequence 7.6 billion people on the planet? This presentation will describe a qualitative methodology to design the required systems for compute and storage as well as the energy usage to sequence 7.6 billion people.

Planting SEEDS to Maximise Data Potential

Margie Smith1, Heather Riley1

1Geoscience Australia, Canberra, Australia, heather.riley@ga.gov.au

 

Geoscience Australia is Australia’s pre-eminent public sector geoscience organisation and the nation’s trusted advisor on geoscience and spatial data.  A key objective for Geoscience Australia is to transform this data into information and knowledge in order to address important national issues, and deliver a broad range of products that assist Government and the community to make informed decisions about the use of natural resources, management of the environment, and community safety.

In order to maintain its reputation as the custodian of geoscientific and spatial data and knowledge, Geoscience Australia must strive to maximise data’s potential. ‘Maximising Data Potential’ is one of the four pillars of the OneGA vision. To achieve this pillar, Geoscience Australia has recently developed and endorsed an enterprise Data Strategy.

The Geoscience Australia Data Strategy was developed through a rigourous consultation process in order to analyse the current state of data at the organisation, and understand the future directions the organisation needs to take to achieve its vision.

From these consultations and analysis it became clear that Geoscience Australia holds a wealth of data and information, but to truly maximise data potential, strong foundations for data and its management need to be laid. The vision for data at Geoscience is for data to be:

  • Accessible
    • Data is open and can be easily retrieved when required
  • Discoverable
    • Data can be found easily and when required
  • Reusable
    • Data can be used again and again, in ways beyond its original intention
  • Managed throughout its lifecycle
    • Data is not managed at a ‘point in time’ during the scientific process; it is part of the scientific process
    • Data is managed to ensure ongoing value can be derived
  • Trusted and quality is well described
    • The strengths and limitations of the data are transparent to users
    • Users have confidence in the data and information Geoscience Australia provides

For Geoscience Australia to successfully maximise data potential, a number of objectives will need to be achieved. These objectives can be referred to as SEEDS

The SEEDS are:

  • Streamline data processes, systems and tools
  • Embed best practice data management
  • Encourage and reward data management
  • Develop data capabilities
  • Strengthen and embed data governance

For a seed to grow and flourish, the right conditions need to be provided. The Geoscience Australia Data Strategy is planting seeds for changes, and providing conditions under which data at Geoscience Australia can grow and flourish so it’s potential can be maximised.

Intersect and La Trobe University – Leveraging a relationship through partnership

Jason Schimming1

1La Trobe University

 

La Trobe University (LTU) joined Intersect as a member in 2016, primarily in response to an increasing need for high-performance computing (HPC) capability in support of La Trobe’s strategic priorities. Over the ensuing three years, the relationship has grown and matured, resulting in LTU expanding its use of HPC, and adopting other major Intersect product and service groups: Time, Energy, Learn, and Space.

This presentation will provide an overview of the development of the relationship between La Trobe University and Intersect Australia, with a focus on three key performance metrics: High Performance Computing, eResearch Analyst initiatives, and realising strategic opportunities.

During the time of La Trobe’s membership, HPC usage has grown strongly, from 75% of the 2016 compute allocation to exceeding 100% of the 2017 allocation and growing further in 2018.

The presence of a full-time, on-campus Intersect eResearch Analyst has been transformative for La Trobe in terms of:

  • Provision of, or the expansion of, eResearch services usage,
  • eResearch skills development encompassing training and expert advice,
  • Researcher consultation and engagement,
  • Identification of strategic opportunities/initiatives and analysis of associated requirements.

As part of the maturing partnership, the timely identification of key technology related pain points and consultative remedial or tactical technical advice, has led to core strategic projects, including a recent project to improve management of the capture and analysis of instrument generated data. Following a detailed requirement gathering and scoping exercise, Intersect worked with LTU to develop a solution that automates the process of capturing data from instruments, upload into Intersect Space, and processing in an Analytical Hub powered by Intersect Time.

La Trobe has a strategic imperative to develop and grow its research capacity, and Intersect play a pivotal role, as a valued partner, in supporting that agenda.


Biography:

Jason Schimming, is the Manager of Infrastructure, Facilities and Storage at La Trobe University, and is also Acting Manager of the Networks group. Jason joined ICT at La Trobe in 2005 and has been involved many and varied technological advances and changes within the university within the IT infrastructure space. In his capacity as Network Manager, Jason is focussed on providing strategic and implementation programs for La Trobe’s storage and data networks that align with the University’s future ready and cloud based strategies

About the conference

eResearch Australasia provides opportunities for delegates to engage, connect, and share their ideas and exemplars concerning new information centric research capabilities, and how information and communication technologies help researchers to collaborate, collect, manage, share, process, analyse, store, find, understand and re-use information.

Conference Managers

Please contact the team at Conference Design with any questions regarding the conference.

© 2018 - 2019 Conference Design Pty Ltd