Home / Faculty / De Los Santos, Jose Aries

De Los Santos, Jose Aries

Lecturer
RoomMath Building Annex 214

Research Interests
Mathematical Optimization, Approximation Theory, Numerical Analysis, Scientific Machine Learning
Academic Groups
  • Education
    • Doctor of Philosophy in Data Science, ongoing
      University of the Philippines Diliman

      Adviser: Dr. Rhudaina Z. Mohammad and Dr. Karel Svadlenka

    • Master of Science in Applied Mathematics (2025)
      University of the Philippines Diliman

      Thesis: Perturbed Accelerated Gradient Method with Adaptive learning rate (PAGMA)
      Adviser: Dr. Renier G. Mendoza and Dr. Rhudaina Z. Mohammad

  • Teaching

    AY: 2025 -2026

    • 1st semester – Mathematics 20  – Precalculus: Functions and Graphs
    • 2nd semester – Mathematics 21  – Elementary Analysis 1

    AY: 2026 -2027 (Mentoring)

    Qualified Scientist/ Technical Adviser (Ongoing) Project: RADAR: Regression-Based Agricultural Drought Analysis and Risk Prediction in Cavite Using Multi-Year NASA POWER Climatological Data, Random Forest, SHAP, and ALE Cavite Science Integrated School (RSHS) – Senior High School (STEM)

    • Provide hands-on mentorship to STEM students in Python programming, applied machine learning (Scikit-learn), and Explainable AI.

    • Lead live coding demonstrations and provide rigorous validation of student-developed predictive models.

    • Guide the application of advanced computational methods to solve local agricultural challenges.

    Qualified Scientist/Technical Adviser (Ongoing) Project: Modified Parameterized Gaussian Error Linear Unit (MP-GELU): Mathematical Analysis and Stable Gradient Propagation in Deep Neural Networks

    Parañaque Science High School – Junior High School (STEM)

    • Provide hands-on mentorship to STEM students in standard calculus, PyTorch, and deep learning systems engineering.

    • Lead the development of baseline implementations (PGELU) and guide the mathematical derivation, asymptotic analysis and implementation of their MPGELU activation function.

    • Supervise architectural design, live coding demonstrations, and of custom CNN training scripts and comparative benchmarking.