Upgrade to Pro — share decks privately, control downloads, hide ads and more …

AI For Good - Machine Learning For Kids - L2

AI For Good - Machine Learning For Kids - L2

AI For Good - Machine Learning For Kids


Nishan Chathuranga

November 21, 2020

More Decks by Nishan Chathuranga

Other Decks in Education


  1. NISHAN CHATHURANGA Software Engineer 99X (Pvt) Ltd. University of Moratuwa

    Faculty of Information Technology WHO AM I
  2. AI FOR GOOD “Whether we are based on carbon or

    on silicon makes no fundamental difference; we should each be treated with appropriate respect.” - Sir Arthur C. Clarke (16 December 1917 – 19 March 2008), 2010: Odyssey Two
  3. AI FOR GOOD Lecture #2



  6. Employment How can AI positively impact the employment rate for

    people with disabilities through more intelligent technology? Daily Life How can AI increase access to technology for people with disabilities, while also decreasing the cost of such technology? Communication & Connection How can AI help improve the speed, accuracy, and convenience of communication for people with disabilities? AGE OF AI Ep3 - youtube.com/watch
  7. Voice-Powered Technology Speech Recognition For someone with severe dyslexia, if

    they’re trying to write down three ideas and they can only speak (to their phone or smart home device), it will make sure all three ideas are jotted down. Smart Home Device Adoption Personal smart devices (Alexa, Google Home, etc), all of them are becoming the norm for people to interact with content. It’s incredibly powerful for people with disabilities. Interactive Voice-Powered Web pages Being able to interact with web pages via voice…there are technologies out there that kind of do it but we haven’t really seen the potential yet
  8. Keyboard Accessibility Someone who only has the use of one

    hand, that’s a much more proper ergonomic option than a one-handed keyboard
  9. Keyboard Accessibility handtrack.js A library for prototyping realtime hand detection

    (bounding box), directly in the browser
  10. https://victordibia.github.io/handtrack.js/#/

  11. Keyboards are our part of life. we use it in

    every computing environment. To reduce our effort in typing most of the keyboards today give advanced prediction facilities. it predicts the next character, or next word or even it can autocomplete the entire sentence. Predictive Keyboards.
  12. Mobile keyboard input is subject to errors that are generally

    attributed to “fat finger typing” (or tracing spatially similar words in glide typing) along with cognitive and motor errors (manifesting in misspellings, character insertions, deletions or swaps, etc). An intelligent keyboard needs to be able to account for these errors and predict the intended words rapidly and accurately. As such, we built a spatial model for Gboard that addresses these errors at the character level, mapping the touch points on the screen to actual keys. The Machine Intelligence Behind Gboard https://ai.googleblog.com/2017/05/the-machine-intelligence-behind- gboard.html
  13. On-device Supermarket Product Recognition One of the greatest challenges faced

    by users who are visually impaired is identifying packaged foods, both in a grocery store and also in their kitchen cupboard at home. This is because many foods share the same packaging, such as boxes, tins, bottles and jars, and only differ in the text and imagery printed on the label. However, the ubiquity of smart mobile devices provides an opportunity to address such challenges using machine learning (ML).
  14. Using “gestures” in different use cases Enabling E-Textile Microinteractions A

    scalable interactive E-textile architecture with embedded touch sensing, gesture recognition and visual feedback.
  15. Using “gestures” in different use cases Collected data from 12

    new participants, which resulted in 864 gesture samples (12 participants performed eight gestures each, repeating nine times), each having 16 features linearly interpolated to 80 observations over time.
  16. Improvements for Translation Gender-specific translations. This feature in Google Translate

    provides options for both feminine and masculine translations when translating queries that are gender-neutral in the source language. For this work, they have developed a three-step approach, which involved detecting gender- neutral queries, generating gender-specific translations and checking for accuracy.
  17. AI for HUMANITARIAN ACTION https://www.sciencedirect.com/science/article/pii /S0956053X19303976#f0080

  18. Use AI-based recovery programs designed for disasters and emergencies According

    to the World Bank, “Artificial intelligence could “end famine” by predicting developing crisis before they begin. The World Bank has launched the Famine Action Mechanism (FAM) in collaboration with international organizations such as the Red Cross, Microsoft etc., to use their expertise and services to prevent famines in the future.
  19. Use AI-based recovery programs designed for disasters and emergencies It

    is often said that in some countries, their legal system are being utilized by governments and influential parties to oppress the voice of people. Courts are being used to punish citizens for ‘crimes’ such as homosexuality and blasphemy. Courts are also being used as instruments of persecution against those who go against the government or demand their human rights. Video: TrialWatch
  20. Improve surgical outcomes of facial surgeries and help more children

    3-Factor Iron Triangle A child with a cured cleft lip condition, post operation With AI gaining prominence in many sectors, the health sector is not far behind. AI has been a transformational factor in changing the lives of many people. Using AI-based services has enabled the medical world to reduce costs and improve the outcomes of treatments. https://news.microsoft.com/transform/operation-smile-dignity-children/
  21. Exploring Nature-Inspired Robot Agility Start by collecting motion capture clips

    of a real dog performing various locomotion skills. Then, we use RL (reinforcement learning) to train a control policy to imitate the dog’s motions. Comparison of policies before and after adaptation on the real robot. Before adaptation, the robot is prone to falling. But after adaptation, the policies are able to more consistently execute the desired skills.
  22. Machine learning programs for vaccine design Start by collecting motion

    capture clips of a real dog performing various locomotion skills. Then, we use RL (reinforcement learning) to train a control policy to imitate the dog’s motions. A diagram of the workflow of MIT's machine learning programs for vaccine design. The OptiVax algorithm searches for optimal binding pairs of peptides and human cell surface receptor proteins. It is composed of a novel assembly of eleven existing machine learning search programs. Its objective function is the information about optimal population coverage fed to it by the second algorithm, EvalVax, which analyzes frequency of genetic variants across the population. Two versions of each program options in the workflow, a simpler version called Unliked and a more sophisticated version, known as Robust, which tracks not just single variants in human genes but linked sets of variants known as haplotypes. The option to cover haplotypes is an advanced feature that sets the search apart from past efforts. Source
  23. None
  24. None
  25. DEMO Image Classification

  26. Transfer Learning Transfer learning (TL) is a research problem in

    machine learning (ML) that focuses on storing knowledge gained while solving one problem and applying it to a different but related problem. For example, knowledge gained while learning to recognize cars could apply when trying to recognize trucks.
  27. MobileNets: Efficient Convolutional Neural Networks MobileNets are based on a

    streamlined architecture that uses depth-wise separable convolutions to build light weight deep neural networks, mainly focuses on mobile and embedded vision applications. In one word the main focus of the model was to increase the efficiency of the network by decreasing the number of parameters by not compromising on performance. READ MORE - medium.com/datadriveninvestor/review-on- mobilenet-v1-abec7888f438
  28. SOURCE CODE GITHUB REPO - github.com/MLforKids BLOG - medium.com/@nishancw/image- classification-with-teachable-machine-ml5-js-and-

    p5-js-233fbdf48fe7 OBJECT DETECTION - github.com/nishanc/ML5jsObjectDetection
  29. FEEDBACK Visit https://www.rateevent.com /speeches/SCAAA038 I would love to hear your

    thoughts or feedback on how I can improve your experience!