title: 【Organized by a Liberal Arts Graduate Who Passed the G Test】36 English Abbreviations for the G Test | CNN, BERT, SHAP… Learn What Each Stands For from Its Meaning in One Go
tags: G Test AI DeepLearning JDLA Certification

Introduction

Hello. I am an engineer from a liberal arts background who passed the G Test in January 2026.

What struck me most when I first began studying for the G Test was, **"There are so many three- to four-letter abbreviations..."** CNN, RNN, GAN, BERT, LIME, SHAP... Trying to memorize them by rote makes them all get jumbled together at once.

So I memorized them using the method of **"expanding each abbreviation and adding a brief explanation of its meaning in Japanese."** Once you know the words in the official name, you can associate them with their roles: "Convolutional, so images" or "Adversarial, so a contest between two networks."

In this article, I organize 36 abbreviations that frequently appear in G Test study by category.

Note: This article is not official content from JDLA (Japan Deep Learning Association). It is an unofficial summary based on the author's personal experience and does not guarantee a pass. Please check the latest syllabus and related materials for the exam's scope.

Memorization Tip: Abbreviations Embed Their Meaning in the "Words Behind the Initials"

Every abbreviation contains a word that can serve as a clue. For example, the following:

RecurrentRecurrent or RepetitiveTime-Series Data (RNN)Competition Between Two Models (GAN)

Word	Meaning	What You Can Associate It With
Convolutional	Convolution	Image Processing (CNN)
	Adversarial	Adversarial
Explainable / Interpretable	Explainable / Interpretable	Shows the reasons for a decision (XAI, LIME)
Retrieval	Retrieval	Retrieval of external information (RAG)

It is perfectly fine even if you feel uncomfortable with English.Instead of following the initials, the trick is to pick out only the “one word that serves as the keyword.”

Abbreviations for Basic Concepts and Learning

Abbreviation	Full Name	Brief Description
AI	Artificial Intelligence	Artificial Intelligence
ML	Machine Learning	A system that learns from data
DL	Deep Learning	Machine learning using a multilayer neural network
MLP	Multi-Layer Perceptron	A multilayer perceptron. The basic neural network
SGD	Stochastic Gradient Descent	Stochastic gradient descent. Updates the model gradually using some of the data
Adam	Adaptive Moment Estimation	An optimization method that adaptively adjusts the learning rate
ReLU	Rectified Linear Unit	An activation function that returns 0 for values 0 or below and returns positive values unchanged
PCA	Principal Component Analysis	Principal component analysis. A representative method for dimensionality reduction
SVM	Support Vector Machine	A method that classifies by maximizing the margin

Abbreviations for Image and Object Detection

Abbreviation	Full name	Brief description
CNN	Convolutional Neural Network	A convolutional neural network for images
ILSVRC	ImageNet Large Scale Visual Recognition Challenge	An image-recognition competition that became famous with the advent of AlexNet
R-CNN	Regions with CNN	Identifies candidate object regions using a CNN
YOLO	You Only Look Once	A fast method for detection in a single pass
SSD	Single Shot MultiBox Detector	A type that performs detection in a single stage
FCN	Fully Convolutional Network	All convolutional layers. The foundation of segmentation
ViT	Vision Transformer	Applies a Transformer to images

Abbreviations for time-series and natural-language processing

Abbreviation	Full name	Brief description
RNN	Recurrent Neural Network	A recurrent neural network for sequential data
LSTM	Long Short-Term Memory	An improved RNN designed to handle long-term memory
GRU	Gated Recurrent Unit	A simplified version of LSTM
NLP	Natural Language Processing	Natural language processing
TF-IDF	Term Frequency–Inverse Document Frequency	A metric for measuring the importance of words
BERT	Bidirectional Encoder Representations from Transformers	A model that understands context in both directions
GPT	Generative Pre-trained Transformer	Predicts the next word and generates text
LLM	Large Language Model	A large-scale language model

Abbreviations related to generative AI, explainable AI, and LLMs

Abbreviation	Full name	In brief
GAN	Generative Adversarial Network	A generator and discriminator compete against each other
VAE	Variational Autoencoder	A variational autoencoder that learns the data distribution and generates data
XAI	Explainable AI	Explainable AI
LIME	Local Interpretable Model-agnostic Explanations	Locally explains individual predictions
SHAP	SHapley Additive exPlanations	Shows the contribution of each feature
Grad-CAM	Gradient-weighted Class Activation Mapping	Visualizes where in an image the model looked to make its judgment
RAG	Retrieval-Augmented Generation	Combines search results to generate an answer
RLHF	Reinforcement Learning from Human Feedback	Reinforcement Learning Using Human Evaluations

Abbreviations Used in Reinforcement Learning and Professional Practice

Abbreviation	Full Name	Brief Description
DQN	Deep Q-Network	A combination of deep learning and reinforcement learning (Q-learning)
MDP	Markov Decision Process	A Markov decision process. The foundation of reinforcement learning
PoC	Proof of Concept	Proof of concept. A small validation before full-scale implementation
CRISP-DM	Cross-Industry Standard Process for Data Mining	A standard approach to carrying out a data analysis project

My Memorization Method

I copied the table above into my notes as-is and worked through it in the following three steps.

1. Hide the abbreviation column and check whether you can give the abbreviation from the full name
2. Hide the “brief description” and check whether you can explain its role from the abbreviation
3. Put sticky notes only on the ones you got wrong and review them just before the exam

Rather than “memorizing everything,”repeating only the ones you got wrongwas more efficient. This is my personal impression, but I also feel that even when an abbreviation suddenly appears on its own in a question, I can now respond without panicking.

Conclusion

Although abbreviations may seem like a high hurdle at first, because so much meaning is packed into the full name itself, I felt that even those with a humanities background could hold their own by **“memorizing from the meaning.”** I hope this table will be at least somewhat useful in organizing everyone’s learning.

Recommended Learning Resources

For those who want to study systematically, we also recommend AI Certification Navi (ai-kentei.net), which compiles study information for AI certifications. It organizes explanations by field and methods for creating study plans, and it also helped me grasp the overall picture of my learning.

Thank you for reading until the end. I hope this helps with your learning!