Skip to content

Chapter 6. Reverse Mode Automatic Differentiation

Reverse mode automatic differentiation computes derivatives by propagating sensitivities backward through a computation. In forward mode, each intermediate value carries a...

SectionTitle
1Chapter 6. Reverse Mode Automatic Differentiation
2Reverse Computational Graphs
3Vector-Jacobian Products
4Reverse Accumulation Algorithms
5Tape-Based Systems
6Wengert Lists
7Memory-Time Tradeoffs
8Checkpointing
9Reverse Mode in Deep Learning
10Case Studies
Chapter 6. Reverse Mode Automatic DifferentiationReverse mode automatic differentiation computes derivatives by propagating sensitivities backward through a computation. In forward mode, each intermediate value carries a...
7 min
Reverse Computational GraphsReverse mode automatic differentiation operates on a computational graph. The forward pass evaluates the graph from inputs to outputs. The reverse pass traverses the same...
6 min
Vector-Jacobian ProductsReverse mode automatic differentiation fundamentally computes vector-Jacobian products. The gradient of a scalar function is a special case of this more general operation.
6 min
Reverse Accumulation AlgorithmsReverse accumulation is the operational core of reverse mode automatic differentiation. The forward pass evaluates a program and records dependency information. The reverse...
6 min
Tape-Based SystemsMost reverse mode automatic differentiation systems require a mechanism for recording the forward computation so that the reverse pass can later traverse it backward. This...
6 min
Wengert ListsA Wengert list is a linear representation of a computation in which every intermediate result is assigned to a unique variable. It is one of the earliest and most influential...
6 min
Memory-Time TradeoffsReverse mode automatic differentiation is computationally efficient for scalar-output functions, but it has a major systems cost: it needs information from the forward pass...
7 min
CheckpointingCheckpointing is a technique for reducing the memory cost of reverse mode automatic differentiation by selectively storing intermediate states and recomputing missing values...
6 min
Reverse Mode in Deep LearningReverse mode automatic differentiation is the mathematical and systems basis of backpropagation. In deep learning, the objective is usually a scalar loss depending on many...
6 min
Case StudiesThis section studies reverse mode automatic differentiation through concrete examples. Each case has the same structure:
6 min