Time Neural ODE (TNODE) Model — TNODE • CellODE

Complete model combining VAE and Neural ODE for cellular dynamics inference. This is the core model of CellODE.

Usage

TNODE(
  n_int,
  n_latent = 5L,
  n_ode_hidden = 25L,
  n_vae_hidden = 128L,
  batch_norm = FALSE,
  ode_method = "euler",
  step_size = NULL,
  alpha_recon_lec = 0.5,
  alpha_recon_lode = 0.5,
  alpha_kl = 1,
  loss_mode = "nb"
)

Arguments

n_int: Number of input features (genes)
n_latent: Dimensionality of latent space (default: 5)
n_ode_hidden: Hidden layer size for ODE function (default: 25)
n_vae_hidden: Hidden layer size for VAE (default: 128)
batch_norm: Whether to include BatchNorm layer (default: FALSE)
ode_method: ODE solver method (default: "euler")
step_size: Step size multiplier for integration (NULL for default)
alpha_recon_lec: Weight for encoder reconstruction loss (default: 0.5)
alpha_recon_lode: Weight for ODE reconstruction loss (default: 0.5)
alpha_kl: Weight for KL divergence (default: 1.0)
loss_mode: Loss function: "mse", "nb", or "zinb" (default: "nb")

Value

nn_module for TNODE model