Attended SLAAI 2024 Conference at KDU, Colombo - Paper Presentation

Tags: Conference, SLAAI, KDU

Presented paper titled “A Review of Automated Bird Sound Recognition and Analysis in the New AI Era”

  • 8th SLAAI - International Conference on Artificial Intelligence 2024
  • Authors Akila Maithripala, Samantha Mathara Arachchi, Kasun Karunanayaka, Ravindu Perera, Pandula Pallewatta

🔗 Presentation Slides 🔗 Presentation Script

Myself At SLAAI 2024 Me presenting

At SLAAI 2024 Receiving presenter certificate from Prof. Yuda

Keynote by Prof Asoka Karunananda

Topic: Using AI to Address Singularity

Prof Asoka Karunananda giving his keynote Prof Karunananda.

Gestation of AI

  • Turing Thesis (1936): Established the concept of computability with six primitives: READ, WRITE, LEFT, RIGHT, DELETE and DO NOTHING to computarize any task.
  • Turing’s Seminal Paper (1952): “Computing Machinery and Intelligence” laid the groundwork for modern AI.
  • John McCarthy’s Proposal (1956): Introduced “Artificial Intelligence” as a distrinct field for intelligent machines. Proposal 🔗
  • Objectives of AI: To understand natural intelligence and build intelligence into machines - AI is the science and engineering of constructing intelligent machines
  • Turing’s Impact: Positioned AI as a software engineering challenge.
  • Neuroscience foundation: AI Draws from neuroscience to explore broad areas of intelligence logic-based and training based- symbolic and non-symbolic AI.
  • Human Evolution: Highlighted the role of symbolic manipulation in humanity’s rapid advancement compared to other species.

Triune of the Brain

  1. Mammal (Peleopallium) - Emotion, Seek Pleasure, Avoid Pain
  2. Reptile (Archipallium) - Survival, Fear
  3. Rational (Neopallium) - Logic and thinking

Intelligence

  • Evolution from Symbolic to Non-Symbolic AI: Transition from Knowledge-driven rule-based systems to data-driven machine learning models
  • LLMs: Blended symbolic and non-symbolic AI to enable creative synergy.
  • Unique Nature of AI: Combines power with control, leading to autonomous decision-making and implications for singularity.
  • Outperforming Human Limitations: AI surpasses humans in situations like,
    • Access time
    • Fatigue, base,
  • AI can solve any problem: This is an implication of the Turing Thesis.

Singularity

  • Achievements of AI: Surpassed expectations, with machines now exceeding human intelligence in various domains.
  • Turing Test Revisited: AI’s ability to “fool the interrogator” is now a reality in many AI solutions - this happens if the interrogator is less knowledgeable about a subject matter.
  • Impacts of Generative AI: Reduces the need for human cognitive skills such as
    • Reading Comprehension, Thinking, Retention, Problem-Solving, Imagination, Communication, Creativity etc.
  • Cognitive Risks: Humanity risks losing hard-earned cognitive capabilities developed over millennia.
  • Acknowledging Risks. Growing awareness of the challenges and dangers associated with singularity.

Facing Singularity

  • Human-Centered AI: Ensures that AI systems are designed with human needs and values at the core - human involvement of final decision making by AI systems.
  • Explainable AI: AI systems go beyond just giving answers, but are able to explain the reasons for the answers - ML solutions still suffer from this issue.
  • Responsible AI: Focuses on the ethical, social and legal aspects of AI to mitigate risks.
  • Tools for cognitive enhancement: develop AI tools to boost cognitive skills such as memory, thinking, language skills and problem-solving skills with humans.
  • Biological Programming: Programming biological system beyond machines modification at the genetic level.
  • Mind Uploading
  • Humanity as Cyborgs: Envisions integration of AI with human biology to augment capabilities and address singularity concerns.

Keynote by Prof Emi Yuda

Japan is facing a super-aging society and there is an urgent need to solve problems in traffic safety technology. She presented their latest research on driver biological condition monitoring technology using bio-signal processing and machine learning.

  • Analysis method for sitting pressure
    1. Pressure distribution mapping (Image processing method)
    2. Time Series Analysis
    3. Frequency Analysis
  • Summary
    • Heart rate variability analysis is useful and particularly excellent for screening for heard disease and sleep disorders.
    • However, conventional methods for estimating fatigue and sleepiness using autonomic estimation methods are not sufficient.
    • They proposed a new analysis method based on sitting pressure analysis which visualized the change in sitting pressure balance over time during a 240 minute sitting simulation operation.

KDU Skyline KDU Skyline.

Back to top