Slide 1

Slide 1 text

Mathematics in the Age of AI and the 4th Generation University Celebration of the π-Day 2025: ”Mathematics, Art, and Creativity” Khemis-Miliana University, March 11, 2025 Prof. Mohammed Hachama, hachamam@gmail.com National Higher School of Mathematics, Algerian Society of Industrial and Applied Mathematics, hachama.github.io/home

Slide 2

Slide 2 text

Outline What is Mathematics? The rise of Artificial Intelligence Recent breakthrough mathematical applications The Fourth Generation University Conclusion Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 2/23

Slide 3

Slide 3 text

AI and Mathematics AI = Teach computers to ”think” like humans Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 3/23

Slide 4

Slide 4 text

AI and Mathematics AI = Teach computers to ”think” like humans Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 3/23

Slide 5

Slide 5 text

AI and Mathematics AI = Teach computers to ”think” like humans Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 3/23

Slide 6

Slide 6 text

AI and Mathematics AI = Teach computers to ”think” like humans Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 3/23

Slide 7

Slide 7 text

AI and Mathematics AI = Teach computers to ”think” like humans y = f (x, P), L = (ytrue − ypredicted )2. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 3/23

Slide 8

Slide 8 text

AI and Mathematics AI = Teach computers to ”think” like humans Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 3/23

Slide 9

Slide 9 text

AI and Mathematics AI = Teach computers to ”think” like humans Very complex optimization problem + data management! Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 3/23

Slide 10

Slide 10 text

Outline What is Mathematics? The rise of Artificial Intelligence Recent breakthrough mathematical applications The Fourth Generation University Conclusion Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 4/23

Slide 11

Slide 11 text

What is Mathematics? What is Mathematics? • Study of numbers, quantities, shapes, structures, and their relationships • Abstraction • Formal language + logical reasoning • Art + Science • Precise tools for understanding reality • Art and Creativity: Inspiring beauty and creativity • Simplicity, Elegance, Conciseness, Symmetry, Unexpected relationships, Infinity, Intrigue. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 5/23

Slide 12

Slide 12 text

What is Mathematics? Key goals of Mathematics • Understanding/Description • comprehend patterns, structures, and relationships. • Application/Calculus • solve real-world problems in science, engineering, and finance. • Optimization • find the most efficient solutions. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 5/23

Slide 13

Slide 13 text

What is Mathematics? Key applications of Mathematics • Engineering • Designing structures, machines, and systems • Finance • Managing investments, risk, and economic models • Cryptography • Securing communication and data • Physics • Modeling natural laws and physical phenomena • Artificial Intelligence • Powering machine learning and data-driven decision-making Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 5/23

Slide 14

Slide 14 text

Outline What is Mathematics? The rise of Artificial Intelligence Recent breakthrough mathematical applications The Fourth Generation University Conclusion Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 6/23

Slide 15

Slide 15 text

Rise of Artificial Intelligence AI, the revolution! • AI = Simulation of human intelligence processes by machines, especially computer systems. • Growth in recent years • Data availability. • Increased computing power. • Development of efficient mathematical models and algorithms, machine learning, and deep learning capabilities. . • Key applications: healthcare, finance, manufacturing, and autonomous systems. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 7/23

Slide 16

Slide 16 text

Mathematical Foundations of AI Mathematics for AI • Linear Algebra: Vector spaces and matrix operations. • Calculus and Optimization: rate of change in learning algorithms. Training AI models via optimization. • Probability and Statistics: making predictions and decisions under uncertainty. Statistical learning. • Logic and Set Theory: Formal reasoning, symbolic AI. • Differential Equations: Modeling in physics-informed AI. Breakthrough Results led to the success of Deep Learning and models like ChatGPT. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 8/23

Slide 17

Slide 17 text

Mathematical Foundations of AI Mathematical Formulation y = f (x, P), L = (ytrue − ypredicted )2. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 8/23

Slide 18

Slide 18 text

Mathematical Foundations of AI Theorem (Universal Approximation Theorem) For any continuous function f : Rn → R defined on a compact subset K ⊂ Rn and for every ϵ > 0, there exists a feedforward neural network function N such that: ∀x ∈ K : |f (x) − N(x)|< ϵ. A NN with can approximate any continuous function. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 8/23

Slide 19

Slide 19 text

How Mathematics Revolutionized AI Attention and transformers • Traditional AI models (RNNs and LSTMs): processed words sequentially, making them slow and inefficient. • In 2017: Google introduced the Transformer, which processes entire sentences at once. • Self-Attention Mechanism: Determines how words relate to each other. Advanced optimization algorithms (such as Adam) + Parallel Computation speed up processing dramatically and made deep learning scalable. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 9/23

Slide 20

Slide 20 text

How Mathematics Revolutionized AI The Age of Large Language Models • GPT-4, ChatGPT, and BERT are all based on Transformers. • Chatbots and Assistants: ChatGPT, Google Bard, Alexa. • Translation: Google Translate uses Transformer-based BERT models. • Search Engines: Google Search ranks pages using BERT. • Medical AI: AI-powered diagnosis and research assistants. • Education and Research • Use of LLM • The role of universities in AI-driven transformation Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 9/23

Slide 21

Slide 21 text

Outline What is Mathematics? The rise of Artificial Intelligence Recent breakthrough mathematical applications The Fourth Generation University Conclusion Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 10/23

Slide 22

Slide 22 text

Recent breakthrough applications Five Innovations 1. Protein Folding: Nobel Prize in Chemistry 2024! 2. First Black Holes images 2019. 3. Cybersecurity: Quantum computers threatening cryptography. 4. Physics-Informed Neural Networks for Weather Forecasting 5. Diffusion Models: A Generative AI Revolution Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 11/23

Slide 23

Slide 23 text

Computational Biology Protein Folding: Nobel Prize in Chemistry 2024! • AlphaFold: Maths + AI to predict 3d shapes of Proteins • Google DeepMind + European Bioinformatics Institute Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 12/23

Slide 24

Slide 24 text

Computational Biology Protein Folding: From 2D sequences to 3D structures • Proteins are fundamental biological molecules. • The function of a protein depends on its 3D shape. • AlphaFold = predicts protein 3D structures from amino-acid sequences using advanced mathematical models. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 12/23

Slide 25

Slide 25 text

Computational Biology Mathematics in Protein Folding • Graph Theory: Representing protein structures. • Neural Networks: Predicting amino acid interactions. • Optimization: Finding the lowest-energy conformation. • Before AlphaFold: • Determining a single protein structure took years. • Required expensive laboratory experiments. • After AlphaFold: • Can predict millions of protein structures in minutes. • Accuracy comparable to lab experiments. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 12/23

Slide 26

Slide 26 text

Computational Biology Revolutionizing Biotechnology and Medicine • Drug Discovery (COVID-19, cancer treatments). • Understanding Genetic Diseases (Alzheimer’s, Parkinson’s). • Synthetic Biology (designing new proteins). • Open database of 200M+ structures (vs. 180k experimentally solved pre-2020). Changing medicine, biotech, and chemistry forever! Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 12/23

Slide 27

Slide 27 text

Mathematics and Black Holes First Image of a Black Hole! • Black holes are invisible, but we can observe their shadow by collecting radio signals. • 2019: This was possible due to mathematics, physics, and AI! Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 13/23

Slide 28

Slide 28 text

Mathematics and Black Holes Black hole imaging = A mathematical problem! • 2019: First image of M87, 55 million light-years away. • 2022: Second image of Sagittarius A, the black hole at the center of our Milky Way (27,000 light-years away). • The imaging process • General Relativity: Einstein’s equations describe black holes. • Fourier Transform: Converts telescope signals into an image. • Inverse Problem Solving: Reconstructs missing data. • Machine Learning: Enhances image clarity. Mathematics continues to push the frontiers of space exploration! Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 13/23

Slide 29

Slide 29 text

Mathematics for Cybersecurity Cryptography • Secure banking, military, financial, internet communications, WhatsApp encryption, Bitcoin, ... • Without cryptography, digital transactions wouldn’t be safe. • RSA and ECC encryption rely on number theory • Factorizing of a number into two prime numbers is hard: N = 15, Factors: 3 × 5 • For a 617-digit number, ”classical” computers take billions of years to factorize. • → RSA works (relies on hard mathematical problems)! Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 14/23

Slide 30

Slide 30 text

Mathematics for Cybersecurity The Quantum Threat to Cryptography • Quantum computers can break RSA using Shor’s Algorithm. • Quantum computers use the Quantum Fourier Transform: |x⟩ → 1 √ N N−1 k=0 e2πixk/N|k⟩ • The world (Governments and industries) is now racing toward post-quantum cryptography. Mathematics is shaping the future of digital security! Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 14/23

Slide 31

Slide 31 text

Physics-Informed Neural Networks (PINNs) The Challenge of Weather Forecasting • Weather forecasting relies on solving Partial Differential Equations (PDEs) that describe atmospheric dynamics. ∂u ∂t + u ∂u ∂x = − 1 ρ ∂p ∂x + ν ∂2u ∂x2 (1) ∂T ∂t + u ∂T ∂x = κ ∂2T ∂x2 (2) • Traditional Numerical Weather Prediction (NWP) models • High computational resources. • Long simulation times. • Large amounts of observational data. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 15/23

Slide 32

Slide 32 text

Physics-Informed Neural Networks (PINNs) Physics-Informed Neural Networks (PINNs) • Traditional neural networks rely only on data. • PINNs incorporate physical laws (PDEs) into the learning process → predictions remain physically consistent even in regions with limited data. • Mathematical Formulation: a combined loss function: L(θ) = Ldata (θ) + λLPDE (θ) • PINNs learn the solution of PDEs directly from sparse or noisy data. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 15/23

Slide 33

Slide 33 text

Physics-Informed Neural Networks (PINNs) How PINNs Improve Weather Forecasting • Faster Predictions (in seconds instead of hours). • Real-time decision-making: agriculture, transportation, and energy. • Less Data Dependency: Can predict in regions with sparse measurements. • Personalized Weather Forecasting: AI-driven hyper-local predictions for cities and industries. • Energy Efficient: Uses a fraction of computational resources. Revolutionizing numerical modeling in other scientific fields. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 15/23

Slide 34

Slide 34 text

Physics-Informed Neural Networks (PINNs) Recent Applications of PINNs in Weather Prediction • FourCastNet (2023): global weather forecasting. • generates a week long forecast in less than 2 seconds • 1000x Faster than traditional numerical models Faster and more reliable hurricane warnings, saving lives and reducing economic damage. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 15/23

Slide 35

Slide 35 text

Diffusion Models: A Generative AI Revolution Generative AI: A Landscape Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 16/23

Slide 36

Slide 36 text

Diffusion Models: A Generative AI Revolution Generative AI: A Landscape • Generative Adversarial Networks (GANs) • Generating realistic images but with some artifacts. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 16/23

Slide 37

Slide 37 text

Diffusion Models: A Generative AI Revolution Generative AI: A Landscape • Variational Autoencoders (VAEs) • Tend to produce blurry images. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 16/23

Slide 38

Slide 38 text

Diffusion Models: A Generative AI Revolution Generative AI: A Landscape • Variational Autoencoders (VAEs) • Tend to produce blurry images. The need for a robust and high-quality generative model. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 16/23

Slide 39

Slide 39 text

Diffusion Models: A Generative AI Revolution The Innovation: Diffusion Models • Formulate image generation as a reverse diffusion process. • Forward SDE: dxt = − 1 2 β(t)xtdt + β(t)dWt • Reverse SDE (learned via score matching): dxt = − 1 2 β(t)xt − β(t)sθ (xt, t) dt + β(t)d ˜ Wt • Generate Data for the ANN. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 16/23

Slide 40

Slide 40 text

Diffusion Models: A Generative AI Revolution Advantages of Diffusion Models • High-quality image generation • avoidance of GANs and VAEs artifacts • Text-to-image synthesis. • Training Data: 10x fewer samples than GANs. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 16/23

Slide 41

Slide 41 text

Diffusion Models: A Generative AI Revolution Faculty of Science and Technology of Khemis-Miliana Univ. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 16/23

Slide 42

Slide 42 text

Outline What is Mathematics? The rise of Artificial Intelligence Recent breakthrough mathematical applications The Fourth Generation University Conclusion Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 17/23

Slide 43

Slide 43 text

University four generations Evolution of University Generations • First Generation: Medieval universities focused on teaching established knowledge. • Second Generation: integrating research alongside teaching. • Third Generation: Entrepreneurial universities integrating innovation and market needs. • Fourth Generation: actively shape their socio-economic environments, emphasizing co-creation and sustainable development. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 18/23

Slide 44

Slide 44 text

University 4.0 Human revolutions • Agricultural Revolution (10,000 BCE): Transition of humans from a nomadic hunter-gatherer lifestyle to a more settled lifestyle based on agriculture and animal husbandry. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 19/23

Slide 45

Slide 45 text

University 4.0 Human revolutions Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 19/23

Slide 46

Slide 46 text

University 4.0 University 4.0/Industry 4.0 Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 19/23

Slide 47

Slide 47 text

Fourth Generation University (4GU) Key Characteristics of 4GU 1. Digital Transformation and Smart Campuses 2. Innovation and Entrepreneurship-driven, and societal Impact 3. Internationalization, global Networks and sustainable development Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 20/23

Slide 48

Slide 48 text

Fourth Generation University (4GU) 1. Digital Transformation and Smart Campuses • Digitalization, AI, Big Data, IoT, and Blockchain. • Blended and personalized learning: AI, virtual labs, and online platforms. • Problem-based learning across multiple disciplines. • Automated university services for student experience improvement. • Smart campuses with digital infrastructure. • Management Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 20/23

Slide 49

Slide 49 text

Fourth Generation University (4GU) 2. Innovation and Entrepreneurship • Encourages technology transfer and startups. • Universities act as incubators and accelerators. • Universities as solution hubs for global issues. • Community-driven research and social engagement. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 20/23

Slide 50

Slide 50 text

Fourth Generation University (4GU) 3. Internat., global Networks and sustainable development • Global collaboration with top universities, research centers, and industries. • Exchange programs and dual degrees to promote international student mobility. • Multilingual education and cross-border learning platforms. • Participation in international research projects on global challenges. • Commitment to the UN Sustainable Development Goals. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 20/23

Slide 51

Slide 51 text

Mathematics in the 4GU Mathematics: The Backbone of Innovation • Shifts in curriculum • Incorporation of AI into mathematics courses. • Emphasis on computational thinking and problem-solving skills. • New teaching methods • Use of AI for personalized learning experiences. • Integration of virtual environments for collaborative projects. • Modeling, Simulation, VR and AR. • Interdisciplinary Applications • Supporting advancements in science, engineering, economics, and beyond. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 21/23

Slide 52

Slide 52 text

Outline What is Mathematics? The rise of Artificial Intelligence Recent breakthrough mathematical applications The Fourth Generation University Conclusion Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 22/23

Slide 53

Slide 53 text

Conclusion and perspectives • Mathematics remains the backbone of Science and Technology advancements, especially in AI and Data Sciences. • Universities must integrate mathematical rigor with AI applications. • Future research should balance theoretical and applied approaches. • Anticipate further integration of mathematics and AI in educational practices. • Collaborative research opportunities between educational institutions and tech companies. Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 23/23

Slide 54

Slide 54 text

Conclusion and perspectives Thank you! Mathematics in the Age of AI and the 4th Generation University M. Hachama (hachamam@gmail.com) 23/23