Hierarchical Cognitron

Closed weights NHK Broadcasting Science Research Laboratories 9.3K parameters April 1984

No estimate

No hardware requirements for this model

The weights for this model have not been published, so it cannot be downloaded or run on your own hardware at any size. It is reachable only through its provider, and no graphics card changes that.

On record

Full specification

Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.

Origin

Who built this model, where, and when it was published.

Organisation
NHK Broadcasting Science Research Laboratories
Organisation type
Industry
Country
Japan
Published
1 April 1984
Authors
K. Fukushima

What it does

The problem areas the model was built for. A model can carry several of each.

Domain
Other
Task
Pattern recognition

Size

How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.

Parameters
9.3K

Parameters 5*5*9*3*3 + 3*3*9*3*3*9 + 9*3*3*9 = 9315 "The numbers of excitatory cells in these four layers were: 7x7 in U0, 5x5 in U1, 3x3 in U2, and 9 in U3" "Each feature-extracting cell in layer U1 receives excitatory modifiable afferent connections from 3x3 cells in layer U0" "On the other hand, each feature, each extracting cell in layers U2 and U3 receives excitatory modifiable connections from all 9 cells in each of the 3 x 3 hypercolumns in the layer preceding it. Therefore, it receives …

Training data
5 tokens

"Five training patterns used for the self-organization are shown in Fig. 4"

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Why it is tracked
Historical significance
Record confidence
Speculative

Sources

Where this record came from and when it was last checked.

Reference
A hierarchical neural network model for associative memory
Last updated
28 November 2025

What the numbers mean

Where it came from

Hierarchical Cognitron was published by NHK Broadcasting Science Research Laboratories, in Japan, in April 1984. industry is the category the publisher falls under.

It works in Other, and is recorded as doing pattern recognition.

Its weights were never published, so it can only be reached through its provider. No graphics card changes that.

How it was trained

The training set ran to roughly 5 tokens.

It is tracked in the underlying dataset for one reason in particular: historical significance.

Answers

Hierarchical Cognitron — common questions

01

What is Hierarchical Cognitron used for?

Hierarchical Cognitron works in Other, and is recorded as handling pattern recognition. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

02

What GPU do I need to run Hierarchical Cognitron?

None. Hierarchical Cognitron is a closed model — its weights were never published, so it cannot be downloaded or run on your own hardware at any price. It is reachable only through its provider.

03

Is Hierarchical Cognitron open source?

The licensing for Hierarchical Cognitron was never recorded in our source data. We treat unstated licensing as closed, because an unrecorded licence is not one to rely on.

04

How many parameters does Hierarchical Cognitron have?

Hierarchical Cognitron has 9.3K parameters. Parameters 5*5*9*3*3 + 3*3*9*3*3*9 + 9*3*3*9 = 9315 "The numbers of excitatory cells in these four layers were: 7x7 in U0, 5x5 in U1, 3x3 in U2, and 9 in U3" "Each feature-extracting cell in layer U1 receives excitatory modifiable afferent connections from 3x3 cells in layer U0" "On the other hand, each feature, each extracting cell in layers U2 and U3 receives excitatory modifiable connections from all 9 cells in each of the 3 x 3 hypercolumns in the layer preceding it. Therefore, it receives 3 x 3 x 9 afferent excitatory modifiable connections altogether". That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.

05

Who created Hierarchical Cognitron?

Hierarchical Cognitron was published by NHK Broadcasting Science Research Laboratories, based in Japan, categorised as industry.

06

When was Hierarchical Cognitron released?

Hierarchical Cognitron was published in April 1984. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

Source

Original publication

Record last updated 28 November 2025

The other direction

Looking at it from the other side?

This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.