Phase 4 · 10 min

Machine learning vocabulary

Terms you will see in every spec

Japanese ML teams mix kanji compounds with katakana loanwords. Learn both: 機械学習 and マシンラーニング are both used, but 機械学習 appears in documents.

EN

Pattern

Most of these compounds are two 2-kanji words glued together. 精度 (accuracy) + 検証 (validation) = 精度検証. 異常 (abnormal) + 検知 (detection) = 異常検知. Once you know 20 building blocks you can read hundreds of terms.

machine learning
deep learning
accuracy validation
feature engineering
anomaly detection
training data
validation data
overfitting
inference
preprocessing
missing values
prediction
classification
regression
production environment

tokuchouryou enjiniaringu de seido ga san pāsento agarimashita

Accuracy went up 3% through feature engineering.

ijoukenchi moderu o honbankankyou ni depuroi shimashita

We deployed the anomaly detection model to production.

kagakushuu o fusegu tameni, kenshou dēta de seidokenshou o okonaimasu

To prevent overfitting, we run accuracy validation on validation data.

Check yourself

1 / 6

13 cards

Add them to spaced repetition

Cards in this lesson (13)

kikaigakushuu

machine learning

shinsougakushuu

deep learning

seidokenshou

accuracy validation

tokuchouryouenjiniaringu

feature engineering

ijoukenchi

anomaly detection

gakushuudēta

training data

kagakushuu

overfitting

suiron

inference

maeshori

preprocessing

kessonchi

missing values

bunrui

classification

kaiki

regression

honbankankyou

production environment

Next lesson

Project and meeting vocabulary

Go