How to Pass the JDLA G Certificate in 2 Weeks

Purpose of this article

The purpose of this article is to share study methods and the benefits of passing the G Test (JDLA Deep Learning For GENERAL) in two weeks.
As prerequisite knowledge, you need to understand high school-level calculus (differential and integral), but if you know that, passing in two weeks is possible.
This is a share of what I caught up on while aiming to become an AI engineer, and I hope it helps you in some way.

Skills gained from certification

The G Test allows you to acquire a wide range of knowledge regarding AI.
The specific contents are as follows:

• History of AI
  You can learn about the transitions in AI booms and the evolution of models, providing background knowledge of AI technology.
• AI Mechanisms
  Within deep learning, you can understand what kinds of models are running for each technology, such as image recognition, object detection, natural language processing, and speech recognition.
• AI Classifications and Types
  You can comprehensively catch up on what models exist for each AI classification, such as supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning.
• Interest in the AI field
  In addition to the above, you can catch up on the laws, regulations, and ethics surrounding AI, which can be a weapon for inexperienced individuals challenging the AI field.

Conversely, the following areas are not covered. I believe these skills can be acquired through the E Certificate (JDLA Deep Learning For ENGINEER), which is a higher-level qualification.

• Coding skills
  No source code appears at all. If you want to improve your coding skills, I recommend challenging certifications like the E Certificate, AI Implementation Exam, or Kaggle.
• Latest AI trends
  While you can catch up on AI history and recent technologies through textbooks, it is difficult to catch up on "hot" technologies that have just emerged.
• Deep mathematical understanding
  While you can understand the overview of how AI works, the specific mathematical formulas driving those mechanisms are in a region that is difficult to understand.

Difficulty Level

The G Test is officially described as follows:

A certification organized by JDLA that certifies individuals with the knowledge and literacy to appropriately utilize AI and deep learning in business.
As the name "Generalist" suggests, it emphasizes business and social implementation perspectives rather than just technology.

I took this exam around the same time as the AI Implementation Exam A-level, and my impression is that the G Test is more difficult due to the breadth of the exam scope and the diversity of how questions are asked.
This exam consists of about 160 questions in 120 minutes. You need to answer each question in 45 seconds, making it a test of whether you deeply understand a wide range of topics.
According to post-exam surveys, the most common average study time is 30 to 50 hours (39%). I believe those who are good at memorization or have practical experience can finish within this timeframe.
Since many questions require combining knowledge to answer and are phrased in slightly tricky ways, it is common to feel uncertain about your answers.
While it is a highly valuable certification for demonstrating interest in the AI field and gaining practical experience, it requires serious effort.

Study Materials

The following are the two textbooks I used.
I mainly used the first one, but the second one contains mock exams that are closer to the actual exam questions. Since both cover the exam content comprehensively, I recommend purchasing the second one.

• [1]Easy-to-Understand Deep Learning G Test (Generalist) Text & Practice (TAC Publishing)
  This textbook is structured so that even those without experience can easily grasp the big picture.
  While it has many problems and is suitable for gaining AI knowledge, it is hard to recommend from the perspective of exam preparation because the questions are too simple and straightforward, which differs from the actual exam trends.

• [2]Strongest Pass Text for Deep Learning G Test (Generalist) [2nd Edition] [Thorough Explanation + High-Quality Problems + Mock Exam (PDF)] (DL Bonus: "This appears on the exam!" Cheat Sheet)
  I borrowed this from a friend the day before the exam and only took the attached PDF mock exam, but it was very helpful as many questions required combining caught-up knowledge to answer, making it very close to the actual exam content. I might have failed if I hadn't solved these problems, so I highly recommend this textbook.

Learning Plan

Broadly speaking, you can obtain the certification with the following flow. I passed in about 60 hours.

Day	Study Content	Study Time
Day 1		1~4h
Day 2	Textbook [1] Chapters 1~4	5h
Day 3	Textbook [1] Chapters 5~8	5h
Day 4	Textbook [1] Chapters 9~12	5h
Day 5	Textbook [1] Chapters 1~4	3h
Day 6	Textbook [1] Chapters 5~8	3h
Day 7	Textbook [1] Chapters 9~12	3h
Day 8	Textbook [1] Chapters 1~6	5h
Day 9	Textbook [1] Chapters 7~12	5h
Day 10	Textbook [1] Comprehensive Review	3h
Day 11	Textbook [1] Comprehensive Review	7h
Day 12	Textbook [1] Comprehensive Review	8h
Day 13	Textbook [2] Mock Exam	5h
Day 14	Actual Exam	2h
Total Study Time		57h

I started without deep knowledge of AI, so I proceeded while understanding each chapter one by one. As a result, I went through Textbook [1] six times and reached a point where I could answer the problems presented with 100% accuracy.
However, even so, my accuracy rate for Textbook [2] problems was barely reaching 70%, so I believe that if you start studying using Textbook [2] from the beginning, you can pass in about 40 hours.

Study Method Utilizing AI

When I study, I make full use of AI (mainly ChatGPT) to catch up.
The purpose of using AI is to verify the correctness of my own hypotheses and gain reproducible knowledge by deepening my understanding.
The main ways I use it are as follows:

• Asking about unknown knowledge
  "What is a determinant?" "The range function sometimes has a different number of arguments, but why?"
  By asking about knowledge seen for the first time, you can quickly get accurate information.
• Asking about relationships between pieces of knowledge
  "What is the relationship between a determinant and an inverse matrix?" "What is the difference between an inner product and a matrix product?" "What are the differences between numpy and pandas?"
  By understanding the connections between knowledge and catching up systematically, you can obtain information that is hard to forget and easy to use.
• Asking about the correctness of your own hypotheses
  "Does a matrix product represent the scaling of a vector?" "Is numpy running behind the scenes of pandas?"
  By posing questions that arise while accumulating knowledge, you can strengthen your memory through output and absorb deeper knowledge.

Exam Day

The G Test is a type of certification that you take from home using your private PC. No camera is required, and you can start the exam from your G Test My Page 10 minutes before the start time.
Due to the large number of questions, you might not have time to review everything. Since you can flag questions you're unsure about, I recommend answering quickly and then thinking deeply during any remaining time.

Secret Tips

The pass rate for this exam is very high (around 80%) relative to its difficulty. I believe the reason is that it's taken at home with the camera off, making cheating possible.
The exam is taken on a dedicated site, and I confirmed that there are no restrictions on taking screenshots. While cheating using books is difficult due to the short exam time, I believe it's possible to send screenshots to an AI to generate answers.
Acquiring a certification is not the goal in itself, but rather acquiring knowledge to use in the field, so I don't recommend cheating.
However, the exam fee is high (13,200 yen), and because the apparent pass rate is high, there is pressure that you "should" pass.
I have personally experienced that I might have failed using only Textbook [1], so if you have studied hard but feel you absolutely won't pass, keep this method in mind.

Summary

The G Test is a significant first step into the AI field.
In the future, AI knowledge will be essential not only for engineers but also for sales and consulting professionals.
Although this certification has a broad scope and a difficult exam, you will gain a lot of knowledge in return, so let's do our best to aim for the certification.