IF2110 - Kecerdasan Artifisial - Artificial Intelligence Overview

Tujuan

• Memahami konsep dasar AI
• Memahami teknologi AI dan sejarah perkembangannya
• Memahami teknologi dan bidang aplikasi AI
• Mengetahui strategi pengembangan AI Huawei
• Mengetahui trend perkembangan AI

AI di Mata Masyarakat

• Berita
• Film
• Aplikasi dalam kehidupan sehari-hari

AI bagi Peneliti

• Alan Turing (1950): Mengajukan pertanyaan, "Can machines think?"
• John McCarthy (1956): Mendefinisikan AI sebagai cabang ilmu komputer yang berkaitan dengan membuat mesin berperilaku seperti manusia.
• Marvin Minsky: Mendefinisikan AI sebagai ilmu membuat mesin melakukan hal-hal yang membutuhkan kecerdasan jika dilakukan oleh manusia.

Cerdas?

• Verbal / Linguistik
• Logis / Matematis
• Visual / Spasial
• Jasmani / Kinestetik
• Musikal / Berirama
• Interpersonal / Sosial
• Intrapersonal / Introspektif / Pribadi

Definisi AI

• Ilmu yang mempelajari dan mengembangkan teori, metode, teknik, dan sistem aplikasi untuk mensimulasikan kecerdasan manusia.

Hubungan AI dengan Bidang Lain

• AI terkait dengan berbagai bidang seperti:
  • Statistik
  • Pengenalan Pola (Pattern Recognition)
  • Neurocomputing
  • Ilmu Otak (Brain science)
  • Ilmu Kognitif (Cognitive science)
  • Ilmu Komputer (Computer science) / Pembelajaran Mesin (Machine Learning)
  • Data Mining
  • Psikologi
  • Filosofi
  • Database
  • KDD (Knowledge Discovery in Databases)
  • Linguistik
  • Logika
• Identifikasi konsep terkait dengan AI dan machine learning.

AI - ML - DL

• Artificial Intelligence (AI)
  • Empat elemen: data, algoritma, computing power, dan skenario
• Machine Learning (ML)
  • Berkembang sejak 1950
• Deep Learning (DL)
  • Berkembang sejak 2010

Kerangka Pemikiran AI

• Simbolisme
• Keterhubungan (Connectionism): Contoh:
  $Apple\ in\ the\ eyes\ of\ connectionism$
  $0.25$
  $0.85$
  $0.31$
  $0.25$
  $0.72$
  $0.65$
  $0.57$
  $0.41$
  $0.32$
  $0.15$
• Tingkah Laku

Sejarah AI

• 1956: AI diusulkan pada Dartmouth Conference.

• 1959: Arthur Samuel mengusulkan machine learning.

• 1976: Karena kegagalan proyek seperti machine translation dan dampak negatif dari beberapa laporan akademis, dana untuk AI berkurang secara umum.

• 1985: Model decision-making tree dengan visualisasi yang lebih baik dan multi-layer ANNs yang menembus batas perseptron awal.

• 1987: Pasar mesin LISP runtuh.

• 1997: Deep Blue mengalahkan juara dunia catur Garry Kasparov.

• 2006: Hinton dan murid-muridnya memulai deep learning.

• 2010: Era big data datang.

• 2014: Microsoft merilis asisten cerdas individual pertama, Microsoft Cortana.

• 2016: AlphaGo mengalahkan juara dunia Go, Lee Sedol dengan skor 4-1.

• 2017: Tim Deep Mind merilis AlphaGo Zero, versi terkuat dari AlphaGo.

• Periode Perkembangan AI

  • 1956-1976: Periode booming pertama. Konsep dan target pengembangan AI ditentukan di konferensi Dartmouth.
  • 1976-1982: Periode low ebb pertama. AI menderita karena pertanyaan dan aturan logika inferensi tidak mampu menjawab pertanyaan spesifik.
  • 1982-1987: Periode booming kedua. Sistem pakar (Expert system) menjadi populer dan komputer generasi kelima dikembangkan.
  • 1987-1997: Periode low ebb kedua. Bidang teknis menghadapi hambatan, orang-orang tidak lagi fokus pada inferensi abstrak, dan model berdasarkan pemrosesan simbol ditolak.
  • 1997-2010: Periode pemulihan. Performa komputasi ditingkatkan dan teknologi Internet dipopulerkan dengan cepat.
  • 2010-Sekarang: Periode pertumbuhan pesat. Teknologi informasi generasi baru memicu transformasi lingkungan informasi dan basis data. Data multi-model seperti gambar, suara, dan teks muncul terus menerus. Kemampuan komputasi ditingkatkan.

Teknologi AI

• Aplikasi
  • Video dan gambar
  • Ucapan suara
  • Teks
  • Kontrol
• Algoritma
  • Machine learning
  • Deep neural network
  • Algoritma optimasi chip
• Perangkat
  • High-bandwidth off-chip memory
  • Bionic device
  • New computing device
  • On-chip memory
  • CMOS process technology
  • New process technology
• Chip
  • Programmable chip
  • Neuromorphic chip
• Proses

Jenis AI

• Strong AI
• Weak AI

Klasifikasi Robot Pintar

• Thinking like human beings
• Thinking rationally
• Acting like human beings
• Acting rationally

Ekosistem Industri AI

• AI application
  • Finance, healthcare, security protection, entertainment, education, retail, agriculture, etc.
• Technical direction
  • Computer vision
  • Speech processing
  • NLP
  • Planning and decision-making system
  • Big data analysis
• AI elements
  • Data
  • Algorithm
  • Computing power
• Basic technologies
  • Big data
  • Cloud computing
  • Server
  • High-performance chip
• Infrastructure
  • Internet, Sensor, and IoT

Sub Bidang AI

• Computer vision
• Natural language processing
• Speech processing
• Machine learning
• Data mining
• Knowledge engineering
• Robot
• Computer Graphics
• Multimedia technology
• Human-computer interaction
• Database technology
• Visualization
• Information retrieval and recommendation

Bidang Teknik & Aplikasi AI

• Areas of AI application
  • Smart healthcare
  • Smart city
  • Smart education
  • Smart finance
  • Digital government
  • Robotics
  • Smart manufacturing
  • Autonomous driving
• Areas of AI technology
  • Knowledge extraction
  • Intelligent adaptive learning
• Research methods
  • Computer vision
  • Planning & Optimizing
  • Expert system
  • Symbolicism
  • Connectionism (e.g. deep learning)
  • Bayesian
  • NLP
  • Backpropagation
  • Probabilistic inference
  • Evolutionism
  • Linear
• Algorithms
  • Genetic programming
  • Inverse deduction
  • Analogizers
  • Decision tree
  • Kernel machine
  • Logistic regression
  • Random forests
  • Support Vector Machine

Arah Penerapan Teknologi AI

• Computer Vision
• Speech Processing
• Natural Language Processing (NLP)

AI Mengubah Semua Industri

• Public sector
  • Safe City
  • Intelligent transport
  • Disaster prediction
• Education
  • Personalization
  • Attention improvement
  • Robot teacher
• Healthcare
  • Early prevention
  • Diagnosis assistance
  • Precision cure
• Media
  • Real-time translation
  • Abstraction
• Pharmacy
  • Fast R&D
  • Precise trial
  • Targeted medicine
• Logistics
  • Routing planning
  • Monitoring
  • Auto sorting
• Finance
  • Doc process.
  • Real-time fraud prevention
  • Up-sell
• Insurance
  • Auto detection
  • Fraud prevention
  • Innovative service
• Retail
  • Staff-less shops
  • Real-time inventory
  • Precise recommendations
• Manufacturing
  • Defect detection
• Telecom
  • Customer service
  • Auto O&M
• Oil and gas
  • Localization
  • Industrial internet
  • Remote maintenance
  • Operation optimization
• Agriculture
  • Fertilization improvement
  • Remote operation
  • Seeds development
• List of industries

AI-Masih Terus Berkembang

• Three Phases of AI

  • Computing intelligence
    • Ability: Capable of storage and computing: Machines can compute and transfer Information as human beings do.
    • Example Value: Distributed computing and neural network
    • Benefits: Help human beings store and quickly process massive data, laying a foundation for perception and cognition.
  • Perceptual intelligence
    • Capable of listening and seeing: Machines can listen and see, make judgments, and take simple actions.
    • Example : Cameras capable of facial recognition and speakers able to understand speeches
    • Benefits: Help human beings efficiently finish work related to listening and seeing.
    • As-is of AI: initial stage of perceptual intelligence
  • Cognitive intelligence
    • Capable of understanding and thinking: Machines can understand, think, and make decisions like human beings.
    • Example: Unmanned vehicles enabling autonomous driving and robots acting autonomously
    • Benefits: Fully assist in or replace partial work of human beings.

Strategi Pengembangan AI - Huawei

• Huawei's Full-Stack, All-Scenario AI Portfolio
  • AI Applications
  • Application Enablement
    • HIAI Engine
    • ModelArts
    • TensorFlow
    • PyTorch Paddle Paddle MindSpore
  • Framework enablement
    • CANN
  • Chip Enablement
    • Ascend-Nano Ascend-Tiny Ascend-Lite
    • Ascend Ascend-Mini Ascend-Max
  • IP & Chip
    • Atlas
• Explanation
  • Application enablement: provides end-to-end services (ModelArts), layered APIs, and pre-integrated solutions.
  • MindSpore: supports the unified training and inference framework that is independent of the device, edge, and cloud.
  • CANN: a chip operator library and highly automated operator development tool.
  • Ascend: provides a series of NPU IPS and chips based on a unified, scalable architecture.
  • Atlas: enables an all-scenario Al infrastructure solution that is oriented to the device, edge, and cloud based on the Ascend series Al processors and various product forms.
• Huawei's "all AI scenarios" indicate different deployment scenarios for AI, including public clouds, private clouds, edge computing in all forms, industrial IoT devices, and consumer devices.

Kontestasi AI

• Bias Algoritme
• Privacy Issues
• Seeing = Believing ?
• AI Berkembang = Pengangguran Meningkat ?
• Mengetahui trend perkembangan AI

Problem

• Produk AI dilindungi Undang-Undang HaKi?
• Siapa yang memberikan wewenang kepada robot ?
• Apa saja hak robot?
• Dst

Masa Depan AI

• Trend Pengembangan Teknologi
• Framework yang lebih mudah dikembangkan
• Performa Algoritme yang lebih baik