Back to Home

Arrow - Artificial Intelligence Experience

Published on

Conference
GitHubLinkedIn

Sponsored Conference by Microsoft, Intel, Nvidia

Arrow - Artificial Intelligence Experience, was a conference that encompassed expert insights into AI, as well as key growth areas in AI that IT professionals must stay updated to.

‘Stay ahead of the AI curve’

Who is Arrow?

Arrow Electronics has guided innovation forward for over 200,000 of the world’s leading manufacturers of technology used in homes, business and daily life.

A technology company that brings innovative IT solutions to market to solve complex business challenges and develops technology solutions that improve business and daily life.

A global provider of products, services, and solutions for customers and suppliers in industrial and commercial markets.

By the time you get your morning coffee, you will have interacted with at least 5 Arrow involved products.

Arrow is involved with everything, it is pretty scary.

Key insights:

  • IOT (Internet of things) is moving very fast to AIOT ( Artificial Intelligent of Things), this is the wave of technology.
  • 80% of companies will increase revenue with IoT -> AIOT, with increased data speeds and availability, AI

Wave of innovation:

  • IOT is a massive market
  • AI growth is accelerating and outpacing IOT.
  • Digital Feedback loops are faster

Data is the new oil

  • Whoever knows how to refine the data will make big money.
  • Data is no good on its own
  • Refined Data = VALUE
  • Act on insights driven from data

Inference Engine

  • A model that has already been trained
  • ready to be published and exported to Edge devices
  • Can use these edge devices with a ready built inference engine, that is hot pluggable.

AI Lifecycle

  • Create
  • Connect
  • Analyse
  • Manage
  • Scale
  • Secure

Microsoft

I'm sure you all have heard of Microsoft: an American multinational technology company. It develops, manufactures, licenses, supports, and sells computer software, consumer electronics, personal computers, and related services.

But as of most recently, Microsoft is heavily invested in the AI space. Microsoft has invested more than $5 billion in IoT + AI. With the key aim to transform industries!

Microsoft want to provide the platform to empower industries + develop new innovative solutions to the world’s most troubling problems. Empower people to transform society. With simple to use Drag and Drop AI models, developers are more ready than ever to innovate without the specific need of a data scientist.

Pace of investment in AI and advanced analytics:

  • 2018: $1.2T
  • 2019: $1.9T
  • 2020: ~$2.6T
  • 2021: ~$3.5T
  • 2022: ~$3.9T
  • This is a REAL industry to get in now! There is a lot of money here!

Microsoft has various AI tools and products available:

  • Azure machine learning studio + workbench
  • Cognitive language and real time voice translation.
  • VSTools for AI
  • Azure Infrastructure:
    • AI On data
    • AI compute
    • CPU, FPGA, GPU
  • AI services
    • Prebuilt AI
    • Cognitive bots

Challenges today:

  • High barriers to entry
  • Massive amount of data we don’t know
  • High costs: Transporting data / cost of latency
  • Limits to act intelligently
  • High developer skill-set needed for hardware.
  • No standardisation in AI - All custom code with manual effort to deploy systems.

Recent AI innovation breakthroughs:

Human parity = Can do the same/ better than humans

  • 2016 - Object detection human parity
  • 2017 - Speech Recognition human parity
  • 2018 - Machine Reading Comprehension human parity - Understanding of the words said.
  • 2019 - Machine Translation human parity - Translation between different languages now real time.

Cognitive Services

Search, Knowledge, language, speech, vision

Microsoft Cognitive Services It is a ready to use service, as well as being very customisable. Vision, Speech, Language comprehension, sentiment detection (if you look happy/sad etc.). knowledge base, search + labs. Easy to use, plug in a containerised REST API.

This solution is incredibly easy to pick up, it is amazing how much power we have access to! Just these services offers a huge amount of use cases, and the possibilities of what humans can do with this tool is endless.

Data at the edge

Cars, smart devices,etc.. are edge devices.

Anything that runs on the edge, should also run on the cloud. Containerised isolated environments allow for services to run on anything. Now we get intelligent clouds.

Security AI at the edge:

  • Mac address theft identification - when a dodgy (Thief) MAC address holder enters a supermarket, there will be an announcement on the tannoy “There is increased security today, in this store” as a preventative mechanism for any thief activity.

There is an Azure Sphere chip: about £6.90. offers 10 year security.

AzureStream analytics:

Performance, Developer productivity, intelligent cloud & the edge. Less code = more developer productivity -> data insights are perishable - insights must be made quickly if you want to make the most impact (Trends -> Figuring out the trends before competitors.) With Azure Solutions I have all the tools needed to really be creative and to start picking up my projects again. I can pick up my skateboarding terrain monitoring project, now I know the tools that can utilise data at the edge.

NVIDIA

Nvidia is also very invested in AI. The company is historically a gaming company, so when people see Nvidia involved in AI conferences there is still quite a bit of confusion. However, did you know that Nvidia is so invested in AI that it is now pretty much 50/50 weighted in gaming and AI.

Amazon’s Reccomendation engine: $86 billion a year

Nvidia in AI

Technology Highlights:

  • Aiva Technologies - Airbus. it is a remote airbus emergency vehicle.
  • Smart Deepsea rover, analyse + improve, maintaineance
    • Wow, this really is my DREAM JOB!
  • Foot waste AI - WINNOW
    • Intelliegent camera to tackle food waste, companies can now be transparent in their food waste.
    • reduce costs in food, better public image, environmental first. Sense of achievement when saving food.
  • BenevolentAI - Healthcare
    • accelerate discoveries with AI -> SUPER RnD
    • AI benefits in Phamacy - reducing time to market.
  • Drone - AerialTonics
    • Aerial inspection, fault detection, dangerous locations, trespassers.
  • enroute
    • Search + Rescue Drones
  • Ocado:
    • predict demand, reduce waste
  • Smart City:
    • Traffic management, improved safety, sensor intelligence
  • Kubernetees on Nvidia GPU cloud:
    • Load balancing, increase magnitude on learning
    • Multi GPU = SUPER LEARNING

AI is just a tool - Not this big thing that will destroy the entire planet.

  • Treat AI as the same as Cloud back in 2010. Everything thought cloud will completly change the world. Well it has had a very large impact, but still just a tool that we can use that provides benefits

Oscar Winners for CGI , all use NVIDIA GPU’s. Rendering Graphics requires billions / trillions compute. AI use cases are very visual right now, and with that, most AI solutions require a GPU.

Autonomous machines (TESLA) the peak in edge devices currently. These vehicles require INSTANT feedback + instant complex decisions. Who to hit.

AI: ITS ALL ABOUT SPEED. ITS ALL ABOUT TIME TO VALUE.

Getting Started

Using AI as a tool to transform the world for the better

Nvidia Jetsonn develpper kits. E105 starter kit. Jetson nano. $100 computers with AI intelligence built in.

Free starter coures to help set up product + produce soem value. Then some additinoal paid courses to take. The field of AI has never been so easy to get into, now their are affordable pathways. - Still is intimidating, I feel like I would need a stronger breadth of tech skill first.

It is important to get started in the Field SOON. All devices WILL BE SMART

Intel

It is the world's largest and highest valued semiconductor manufacturer based on revenue, the processors found in most personal computers (PCs). Intel is playing a huge part in AI, creating the hardware built for AI. Branding their latest hardware solutions with 'Intel Deep learning boost', they sure are finding the money in AI.

OPEN VINO

Leveraging powerful neural networks processing video for ai applications as fast as possible. Microsoft and intel have been working together lots to work on building a great AI experience. OpenVino provides lots of compute power. Optimising the data before data manipulation - removing useless data that will not affect AI performance. Only speed it up.

Open Vino Model Zoo:

  • Loads of pre trained models: face analysis, pretrained DL models, best performing solution across all intel based architectures.
  • Training Phase -> Inference phase.
  • Increased speed of development: Reduced time by using a library of optimised solutions.

DL Workflow

  1. Import a pretrained model - Eliminate unneccesary layers
  2. use the model - immediate representation
  3. Inference engine - optimised CROSS platform

Customer information

There is a lot you can do without specific customer PI information.

  • Retail decisions: Complexion - See what clothing a customer suits when entering the store, automated reccomedation.

Examples:

Video indexer:

  • Summarise keynote presentations very easily. Only hear what you want to hear with a selection of tags that match voice records.
  • One way of doing speed learning. Cut out the waffle.
  • Can apply to recorded lecture recordings!
  • Or more applicable, extremely long webinars, that you just need to get succint information from.
  • Video content -> Written transcript, that is indexable.

Custom Vision

  • Sentiment analysis: detect the mood the user is in.

Thanks

Thank you for reading this blog post. This was a long one! (for me anyway) It took me a while to write up this one properly, so unfortunately content may not be as clear. My aim for next time is to write up my learnings whilst they are still fresh in my head, and I am full on inspiration and ideas!