NCA-GENM - Experimentation - Section 1.2

Improve image quality with denoising diffusion processes.

Explain how diffusion models progressively add and then reverse noise to learn a data distribution and generate sharp images. Distinguish the effect of scheduler choice, guidance scale, and step count on output fidelity.

Diffusion models

Practice question for this objective

Free sampleExperimentationhard

During the forward diffusion process of a DDPM, a practitioner wants the model to spend proportionally more training time on moderately noisy images rather than near-pure-noise states. Which noise schedule property best achieves this?

  • AA cosine noise schedule defined by the signal-to-noise ratio following a cosine curve, which keeps images perceptually structured for more timesteps before rapid noise saturation. Correct
  • BA linear beta schedule that increments the noise variance by a constant amount at each timestep from a small starting value to a maximum near one.
  • CA quadratic beta schedule that increases variance slowly at first and rapidly at the end, compressing most variance growth into the final timesteps.
  • DA sigmoid beta schedule that applies maximum variance in the early timesteps and tapers off, ensuring clean images receive the greatest denoising signal.
Explain how the cosine noise schedule controls the signal-to-noise ratio trajectory to improve training coverage of intermediate noise levels in diffusion models. In DDPM training, the model learns to reverse each step of a forward process that progressively adds Gaussian noise. With a linear beta schedule, the signal-to-noise ratio collapses to near zero quickly, leaving many timesteps where the image is essentially pure noise and the denoising task provides little useful gradient. The cosine schedule (Nichol and Dhariwal, 2021) defines alpha-bar as a cosine function of the timestep fraction so that the signal decays slowly through the intermediate range and saturates only near T. This keeps a larger fraction of training timesteps in the perceptually informative zone, improving the learned denoising prior for natural images and producing higher-quality samples.

Why A is correct: The cosine schedule, introduced to address linear schedule deficiencies, defines alpha-bar values along a cosine curve so the signal-to-noise ratio decays slowly through intermediate timesteps and only reaches near-zero close to T, naturally concentrating training effort on the perceptually rich moderate-noise regime.

Why B is wrong: A linear schedule allocates equal variance increments across all timesteps, meaning it devotes as many steps to the near-pure-noise end as to the structured intermediate range, which does not concentrate training on moderately noisy images.

Why C is wrong: A quadratic schedule accelerates variance growth toward the end of the chain, which concentrates steps near the heavily corrupted end rather than in the intermediate regime, worsening the imbalance the practitioner wants to correct.

Why D is wrong: Applying maximum variance early would corrupt clean images aggressively in the first few steps, destroying semantic structure before the model has learned to handle it, and does not correspond to any established noise schedule used in diffusion model literature.

See more NCA-GENM practice questions, answers explained.

More in this domain

Back to all Experimentation objectives, or the NCA-GENM cert hub.

Examworthy is not affiliated with or endorsed by NVIDIA. Original, blueprint-aligned practice material only.