Graduate ML Pre-Study Plan (CSCE 5215)

Timeline:
Today: Dec 29, 2025
First day of class: Mon, Jan 12, 2026
Schedule: Full-time work (Mon–Fri), Jan 1 off

Guiding Principles

• Recognition > mastery before class
• Classical ML first, deep learning second
• One focused task per day
• Avoid burnout before Week 1

Mon Dec 29 – Wed Dec 31 (Workdays)

Theme: Conceptual deep learning primer (45–75 min/day)

• Deep Learning with Python (Chollet)
  • Finish Chapter 1
  • Begin Chapter 2 (math building blocks)
  • Focus on tensors, loss functions, gradients (conceptual)
  • Skip heavy derivations if needed
• Optional (one day only):
  • StatQuest: Gradient Descent
  • or 3Blue1Brown: Neural Networks (Part 1)

Thu Jan 1 (Day Off)

Theme: High-leverage consolidation

• Deep Learning with Python
  • Finish Chapter 2
  • Skim Chapter 4 (training workflow, no coding)
• Hands-On Machine Learning (Géron)
  • Read Chapter 1: The Machine Learning Landscape (fully)

Fri Jan 2 (Workday)

Theme: ML workflow framing

• Hands-On ML (Géron)
  • Skim Chapter 2: End-to-End ML Project
  • Focus on data → model → evaluation flow
  • Ignore code details

Sat Jan 3 – Sun Jan 4 (Optional)

Theme: Light reinforcement or rest

• Re-read Géron Chapter 1 notes
• or Watch one StatQuest video (bias–variance or overfitting)
• or Rest completely

Mon Jan 5 – Wed Jan 7 (Workdays)

Theme: Core classical machine learning

• Hands-On ML (Géron)
  • Chapter 3: Classification
  • Focus on decision boundaries, overfitting, evaluation metrics
  • Chapter 4: Training Models (gradient descent, regularization)

Thu Jan 8 – Fri Jan 9 (Workdays)

Theme: Probabilistic exposure (light)

• Probabilistic Machine Learning (Murphy)
  • Read Chapter 1 only
  • Goal: notation familiarity and probabilistic framing
  • Optional skim of probability review if rusty

Sat Jan 10 – Sun Jan 11

Theme: Calm confidence

• Review personal notes and summaries
• Watch one intuition video if desired
• Skim Géron Chapter 1 headings
• Avoid: cramming, coding, heavy math

Mon Jan 12 — First Day of Class

• Stop pre-studying
• Arrive rested and oriented
• Use Murphy during lectures, not before them

Expected Outcome by Day 1

• Clear intuition for models, loss, training, and generalization
• Immediate recognition of ML terminology
• Reduced cognitive load during Weeks 1–2
• Strong foundation without burnout