1 Automating Machine Learning and Deep Learning Workflows 2 - - PowerPoint PPT Presentation

▶
1 automating machine learning and deep learning workflows
SMART_READER_LITE
LIVE PREVIEW

1 Automating Machine Learning and Deep Learning Workflows 2 - - PowerPoint PPT Presentation

Jan 07, 2023 546 likes •798 views

1 Automating Machine Learning and Deep Learning Workflows 2 Information Name: Mourad Mourafiq Author of an open source platform: Polyaxon twitter: @mmourafiq GitHub: mouradmourafiq 3 What is Polyaxon Solves the

slide-1
SLIDE 1

1

slide-2
SLIDE 2

Automating Machine Learning and Deep Learning Workflows

2

slide-3
SLIDE 3

Information

  • Name: Mourad Mourafiq
  • Author of an open source platform: Polyaxon
  • twitter: @mmourafiq
  • GitHub: mouradmourafiq

3

slide-4
SLIDE 4

What is Polyaxon

  • Solves the machine learning life cycle
  • Can be deployed on premise or on any cloud

platform

  • Is open source
  • Works with any library or framework
  • Can be used by single users or large organizations
  • Provides compliance, auditing, and security

4

slide-5
SLIDE 5

Why you need a tool to manage your ML

  • perations?
  • Software development is mature
  • Why not use the same tools?
  • What is the difference between software

development and ML development?

  • What is the difference between software

deployment and ML deployment?

5

slide-6
SLIDE 6

Difference between software development and ML development

  • Development objectives
  • Vetting and quality assurance
  • Development stack

6

slide-7
SLIDE 7

Difference between software deployment and ML deployment

  • ML deployment needs a Feedback Loop
  • Iteration and refinement
  • People involved in the deployment cycle

7

slide-8
SLIDE 8

What should a ML platform answer

  • Should be flexible to support open source initiatives
  • Provides different deployment options
  • Ideally open source
  • Works with any library or framework
  • Scales with users
  • Provides compliance, auditing, and security

8

slide-9
SLIDE 9

ML development lifecycle

  • Data access
  • Data exploration and Feature engineering
  • Experimentation: iteration, packaging, reusability, reproducibility.
  • Scaling: Scheduling, orchestration and optimization
  • Tracking: code, data, params, artifacts, metrics
  • Insights, reporting, and knowledge distribution
  • Model management: packaging, deployment, and distribution
  • Compliance, auditing, and access management.
  • Automation, events, and workflows
  • User experience

9

slide-10
SLIDE 10
  • Data access

10

slide-11
SLIDE 11
  • Data exploration & Feature engineering

11

slide-12
SLIDE 12
  • Experimentation
  • Different environments: local, remote, cluster
  • Portability and reusability
  • Reproducibility

12

slide-13
SLIDE 13
  • Experimentation: Different environments

13

slide-14
SLIDE 14
  • Experimentation: Packaging
  • polyaxon run -f polyxonfile.yaml
  • polyaxon run -f polyxonfile.yaml —local

14

slide-15
SLIDE 15
  • Scheduling & Orchestration

15

slide-16
SLIDE 16
  • Hyperparams tuning & distributed training

16

slide-17
SLIDE 17
  • Experiments tracking

17

slide-18
SLIDE 18
  • Experiments tracking

18

slide-19
SLIDE 19
  • Insights, reporting, and knowledge distribution

19

slide-20
SLIDE 20
  • Model Management

20

slide-21
SLIDE 21
  • Compliance & Governance
  • Manage model development and deployment
  • Rigorous and auditable workflows

21

slide-22
SLIDE 22
  • Automation & Events
  • Simple yet effective specification to create

workflows and automation

  • Integration with other pipelining tools, e.g.

airflow

  • Events and triggers based on data, code,

metrics, …

22

slide-23
SLIDE 23

mourad@polyaxon.com twitter: @mmourafiq GitHub: mouradmourafiq https://polyaxon.com twitter: @polyaxonai GitHub: polyaxon

23