Bo Zheng
  • blog (current)
  • about
  • CV
  • flappy
Thesis
  • Part3.5-PERecoveryProof
  • Part2.5-ASharpGeneralizationBound
  • Part5-PENHolonomyandSingleTangentFallacy
  • Part4-CategoryTheoryPerspective
  • Part2-GroupStructure
  • Part3-PathEquivariance
  • Part1-PriorKnowledge
All Articles
  • PML-1 MAP MLE KL
  • TheWorldFromWithinAndWithout
  • From Distances to Coordinates (Euclidean)
  • PML-2 From Likelihood to ELBO
  • Independence in Bayesian Network Causal Diagrams
  • Semidefinite Programming and Applications
  • Yoneda Perspective
  • Pushforward Pullback
  • What is dx?
  • Category Product
  • Product
  • Group Ring Field
  • Monoid
  • Paradox
  • PCA
  • Probability-0
  • Part2.5-ASharpGeneralizationBound

    Orbit directions are trivially flat, inflating sharpness estimates. Quotient-space sharpness factors out reparametrization symmetry for tighter generalization bounds.

    3 min read   ·   April 28, 2026   ·   0

    2026   ·   math   machine-learning   thesis   deep-learning

  • Part3.5-PERecoveryProof

    A complete proof that classical group equivariance is recovered from path equivariance under the endpoint condition, establishing classical equivariant networks as a special case of the PEN framework.

    12 min read   ·   April 28, 2026   ·   0

    2026   ·   math   machine-learning   thesis   deep-learning   differential-geometry

  • PML-1 MAP MLE KL

    MAP vs MLE as point estimates, global parameters vs per-example latent variables, the ELBO, and why the two directions of KL divergence give fundamentally different approximations.

    6 min read   ·   April 28, 2026   ·   0

    2026   ·   math   machine-learning   probability

  • Part5-PENHolonomyandSingleTangentFallacy

    Path equivariant networks via parallel transport, holonomy-controlled expressivity, and why the single tangent space approach fails on curved manifolds.

    14 min read   ·   March 31, 2026   ·   0

    2026   ·   math   machine-learning   thesis   deep-learning   differential-geometry

  • TheWorldFromWithinAndWithout

    Intrinsic and extrinsic perspectives in mathematics, physics, and philosophy.

    8 min read   ·   March 31, 2026   ·   0

    2026   ·   math   differential-geometry   philosophy

  • Semidefinite Programming and Applications

    SDP formulation, duality, and applications to Euclidean distance completion and sparse PCA.

    11 min read   ·   March 26, 2026   ·   0

    2026   ·   math   optimization

  • Independence in Bayesian Network Causal Diagrams

    Independence and conditional independence in Bayesian networks, d-separation, and the collider effect.

    10 min read   ·   March 26, 2026   ·   0

    2026   ·   math   probability   machine-learning

  • PML-2 From Likelihood to ELBO

    The probabilistic ML pipeline: notation, likelihood, ELBO derivation, and the reparameterization trick for VAEs.

    20 min read   ·   March 26, 2026   ·   0

    2026   ·   math   machine-learning   probability

  • From Distances to Coordinates (Euclidean)

    Recovering point coordinates from pairwise distances via Gram matrices and eigendecomposition.

    5 min read   ·   March 26, 2026   ·   0

    2026   ·   math   linear-algebra

  • Yoneda Perspective

    Understanding the Yoneda Lemma and its deep implications in category theory.

    5 min read   ·   March 25, 2026   ·   0

    2026   ·   math

  • Part1-PriorKnowledge

    Prior knowledge in neural networks: every design choice encodes a structural assumption about the world.

    4 min read   ·   March 25, 2026   ·   0

    2026   ·   math   machine-learning   thesis   deep-learning

  • Part3-PathEquivariance

    Generalizing group equivariance to path equivariance on manifolds, with fiber bundles and the content-pose decomposition.

    15 min read   ·   March 25, 2026   ·   0

    2026   ·   math   machine-learning   thesis   deep-learning   differential-geometry

  • Part2-GroupStructure

    How activation functions and regularization break the symmetry group of deep networks, traced from GL(n) through specific subgroups.

    20 min read   ·   March 25, 2026   ·   0

    2026   ·   math   machine-learning   thesis   deep-learning   algebra

  • Part4-CategoryTheoryPerspective

    Equivariance is naturality: unifying groups, manifolds, and path equivariance through category theory.

    12 min read   ·   March 25, 2026   ·   0

    2026   ·   math   machine-learning   thesis   deep-learning   category-theory

  • Pushforward Pullback

    Pushforward and pullback in differential geometry and probability, with the duality between vectors and forms.

    25 min read   ·   February 21, 2026   ·   0

    2026   ·   math   differential-geometry

  • What is dx?

    What dx means: from calculus infinitesimals to differential 1-forms on manifolds.

    7 min read   ·   January 07, 2026   ·   0

    2026   ·   math   differential-geometry

  • Category Product

    The categorical product via universal property, with arguments about arrow direction and coproducts.

    4 min read   ·   November 01, 2025   ·   0

    2025   ·   math   category-theory

  • Product

    Products across mathematics: inner/outer/cross products, Kronecker, group products, tensor products, and their universal properties.

    10 min read   ·   October 05, 2025   ·   0

    2025   ·   math   algebra   category-theory   linear-algebra

  • Probability-0

    Probability spaces, conditional probability, random variables, independence, and expectation from measure theory.

    11 min read   ·   October 04, 2025   ·   0

    2025   ·   math   probability

  • PCA

    PCA as eigendecomposition of the covariance matrix, its SVD implementation, and why it works.

    6 min read   ·   October 04, 2025   ·   0

    2025   ·   math   machine-learning   linear-algebra

  • <
  • 1
  • 2
  • >
© Copyright 2026 Bo Zheng. Powered by Jekyll with al-folio theme. Hosted by GitHub Pages.