T1913

Ladda ner som PDF

Architecture for AI and Machine Learning

Whether you’re just starting to keep an eye on ML, or are already adjusting your architecture to the current wave of AI, this course addresses several challenges most architects will be facing in an architectural landscape where ML is a fast-grower. The course is not tied to a particular vendor, language, framework, library, or environment.

Remember the architect’s tightrope walk between flexibility/performance and testability/reliability/predictability/V&V? For example, due to run-time parameters, or parallelism, or late binding and SOA’s UDDI…  Right now, Machine Learning (ML) is leveling up the same tradeoff.

Audience

- Architects, designers, modelers, analysts, tech managers and experts involved in (or affected by) AI projects.

Prior knowledge

Agile Architecture Fundamentals (course T1101) or corresponding experience/knowledge of IT architecture in an agile setup. Even experience from other roles in the IT field is useful, especially from knowledge-intensive sectors.

Courseware

Presentation slides and short exercise, in English.

Course Outline: 

Introduction:  Abstraction  levels, AI, ML’s past and present

  • Abstraction levels: architect versus devs
  • AI overview: pushing up the bar for intelligent systems
  • The evolution of ML:  from slow and auditable, to fast smart and opaque. Despite ML’s roots (rule induction and mining) in the successful decryption of Wehrmacht’s unbreakable cipher Tunny, its near future seems encoded as weight values, in deep neural networks. Whereas (for example) black-box flight recorders clarified the chain of events & decisions in past emergencies, ML has step-by-step landed in black boxes that hide opaque logic.
  • Predictions for the next 10-12 months: AI investment, and tech challenges to cope with.

Consequences across the landscape:  data architecture, SWA, analytics, HMI, RT

  • NFR (non-functional/“quality” requirements during the move toward business apps and mission-critical; in most sectors of industry, “better on average” is no longer good enough (unlike in marketing or customized offers for billions of layman consumers)
  • Three categories of ML
  • Deep Learning.

The role of architects

  • What is, and is not, in the job description of an architect
  • Perspective (broad yet integrated) compared to researcher’s or prime mover’s (focused, deep, yet often green-lawn biased)
  • The systems... Läs mer

Course Outline: 

Introduction:  Abstraction  levels, AI, ML’s past and present

  • Abstraction levels: architect versus devs
  • AI overview: pushing up the bar for intelligent systems
  • The evolution of ML:  from slow and auditable, to fast smart and opaque. Despite ML’s roots (rule induction and mining) in the successful decryption of Wehrmacht’s unbreakable cipher Tunny, its near future seems encoded as weight values, in deep neural networks. Whereas (for example) black-box flight recorders clarified the chain of events & decisions in past emergencies, ML has step-by-step landed in black boxes that hide opaque logic.
  • Predictions for the next 10-12 months: AI investment, and tech challenges to cope with.

Consequences across the landscape:  data architecture, SWA, analytics, HMI, RT

  • NFR (non-functional/“quality” requirements during the move toward business apps and mission-critical; in most sectors of industry, “better on average” is no longer good enough (unlike in marketing or customized offers for billions of layman consumers)
  • Three categories of ML
  • Deep Learning.

The role of architects

  • What is, and is not, in the job description of an architect
  • Perspective (broad yet integrated) compared to researcher’s or prime mover’s (focused, deep, yet often green-lawn biased)
  • The systems portfolio and digital twins (“avatars”) cross-connecting the two.

Tackling the leveled-up tradeoff

  • Top-down: a framework of corporate values and procedures to boost V&V and transparency
  • Bottom-up: tech and practices in IT to build in transparency mechanisms upfront (visualization, explanation, traceability, etc.)

Explainable AI

  • “Big bang,” via research into self-explanation capabilities for deep neural networks
  • Corporate “think big, start small,” via hybrid AI.

End of the course 

About Teacher

Milan Kratochvil, teacher Informator UtbildningMilan Kratochvil , senior consultant, Kiseldalen.com (Sweden). Main author: UML Extra Light (Cambridge University Press) and Growing Modular (Springer).

Since the mid 1990-ies, he also cooperates with Informator-Tieturi, Scandinavia’s largest training provider for IT-professionals.

Utbildningen levereras i samarbete med

Kursfakta

Kurs-ID: T1913
Längd: 1 dag
Pris exkl moms: 9 900 kr
Inregistrering: 09.00
Kursstart: 09.30
Kursslut (ca): 17.00

Frågor om kursen?

Har du frågor om kursens innehåll, leveransdatum/ort eller behöver en företagsanpassad variant? Fyll i formuläret nedan!


Kan betalas med:
TRAINING CARD

Ort och datum

Stockholm
15 mar
Boka nu!
15 apr
Boka nu!

Tipsa