Quantization

As explained in Math and Architectures of Deep Learning

The conversion of floating-point region boundaries or ratios into integer-aligned indices for feature extraction.

Quantization converts continuous or floating-point values into discrete buckets or integer-aligned indices. In feature extraction, it makes RoI pooling regions executable on a discrete feature map. section 5.6 section 11.4.3

The treatment uses quantization in two related ways:

• Discretize random variables by assigning ranges such as weight or height intervals to buckets and estimating each bucket’s probability. section 5.4.2
• Convert floating-point RoI-to-bin ratios into integer region sizes, using the closest integer or floor operation in the original implementation. section 11.4.3
• Preserve a fixed output shape while allowing the input RoI and pooling-region size to vary. section 11.4.3

Quantizing an RoI for pooling

If a feature-map RoI has height 8 and the target pooling grid has four rows, each region is 8/4 = 2 feature-map units. When the ratio is fractional, quantization selects an integer size; RoIAlign instead uses bilinear interpolation. section 11.4.3

Figure 11.18 A conv feature map with two regions of interest of different sizes. RoI pooling extracts a fixed-sized output feature map (2 × 2 in this image) from each of the RoIs in a single pass via max pooling. This enables us to extract fixed-sized representative feature vectors for each RoI, which are then fed into further classifier and regressor layers for classification and localization.

Compare differently sized RoIs producing fixed-sized pooled feature maps through max pooling. section 11.4.3

How Math and Architectures of Deep Learning covers it

How deep it goes

Defines bucketization and integer rounding, explains their effects on probability distributions and RoI pooling, and defers RoIAlign details.

Examples

Uses height and weight buckets, fractional RoI dimensions, and fixed-size pooled feature maps as concrete examples.

What it assumes

Assumes basic probability and feature-map geometry while explaining bucketization and floating-point-to-integer conversion directly.

What you can do after

Apply it via RoI pooling to turn variable-sized regions into fixed-sized feature representations.

How it connects

Connects quantization with random variables, probability distributions, RoI pooling, max pooling, RoIAlign, and bilinear interpolation.

liveBookmanning.com

Read Math and Architectures of Deep Learning on liveBook Get the book at manning.com

Also explained in

• 

  liveBookmanning.com
  LLMs in ProductionQuantization is the conversion of model parameters or computations from higher-precision numerical representations to lower-precision representations.
• 

  liveBookmanning.com
  Domain-Specific Small Language ModelsMethods that reduce numeric precision or representational size of model values to improve efficiency, especially for inference; here the absmax method is a specific quantization procedure.
• 

  liveBookmanning.com
  AI-Powered SearchQuantization is the transformation of numerical values into a smaller set of discrete values to reduce representation precision and resource requirements.
• 

  liveBookmanning.com
  Deep Learning with Python, Third EditionThe technique of reducing numeric precision of model weights or activations to speed up computation while preserving output behavior as much as possible.
• 

  liveBookmanning.com
  Deep Learning with R, Third EditionModel quantization is the conversion of model computations or parameters to lower-precision numerical representations to reduce resource requirements.
• 

  liveBookmanning.com
  Generative AI in ActionQuantization is the reduction of numerical precision used to represent model parameters, typically to reduce memory use and computational cost.
• 

  liveBookmanning.com
  Deep Learning with PyTorch, Second EditionA model optimization technique that reduces numerical precision of parameters and operations to lower memory use and often speed up inference.
• 

  liveBookmanning.com
  Transformers in ActionQuantization is the transformation of numerical model parameters or activations into lower-precision representations to reduce memory consumption or computational cost.
• 

  liveBookmanning.com
  The Complete Obsolete Guide to Generative AIA quantized model is a machine-learning model whose numerical parameters or activations are represented at reduced precision to lower resource requirements.
• 

  liveBookmanning.com
  Time Series Forecasting Using Foundation ModelsQuantization is the conversion of continuous numerical values into a finite set of discrete levels or tokens.
• 

  liveBookmanning.com
  Natural Language Processing in Action, Second EditionQuantization is the transformation of numerical values into lower-precision discrete representations, often to reduce memory use and accelerate computation.
• 

  liveBookmanning.com
  Building Reliable AI SystemsQuantization reduces the numerical precision used to represent model values, often lowering storage or computation requirements.

Compare how 13 books explain quantization →

Related concepts

• QLoRAused by · QLoRA is a parameter-efficient fine-tuning method that quantizes a pretrained model’s weights while training low-rank adapters to reduce memory requirements.
• scalar quantizationincludes · Scalar quantization maps numerical scalar values to a finite set of discrete levels, typically reducing their precision or range.
• QA-LoRAused by · QA-LoRA is a low-rank adaptation method that integrates quantization-aware operations and group-wise processing to support efficient low-bit adaptation.
• post-training quantizationincludes · Post-training quantization is the conversion of a trained model’s parameters or computations to lower-precision representations without additional training.
• quantization-aware trainingincludes · Quantization-aware training optimizes a model while simulating or incorporating reduced-precision effects so it adapts to quantized representations.
• PyTorchuses · PyTorch is an open-source machine-learning framework for Python that provides tensor computation, automatic differentiation, and tools for building and training neural networks.

This explainer was generated by Manning's knowledge graph from the text of Math and Architectures of Deep Learning. The section links point to the parts of the book it draws on.