本作品選自張晨(Chen Chang)老師開設的「學術英文口頭報告」課程。該課程專為研究生設計，旨在培養學術口頭簡報技巧，涵蓋針對不同聽眾設計簡報、善用視聽輔助器材、訓練聲音技巧，以及運用說服語言策略等核心能力。張晨老師具有國立中正大學及塔夫斯大學雙碩士學位，專精於學術英文寫作發表與口頭報告教學。

Turning Research into Impact

    Communicating complex research in just three minutes is a demanding task that calls for more than your technical knowledge. It requires clarity, creativity, and deep audience-awareness. The 3-Minute Thesis (3MT) competition challenges researchers to present their work in a clear and engaging way that can be easily understood by a general audience, regardless of their field or expertise. It is an exercise in not only explaining what you have discovered, but showing why it matters.

    This semester, the 8th 3MT competition offers that opportunity once again, with the final event scheduled on Thursday, November 13. As you get ready to take the stage, an effective way to prepare is by studying good examples. The following student script, titled“Your Car's Secret Bodyguard,” showcases a successful approach to a technical subject, namely“Federated Learning”in connected vehicles. The analysis following the script highlights its strengths and shows how you can incorporate them into your own presentation.

Your Car's Secret Bodyguard
電機所 鄧雅文

    Imagine you're wearing a smartwatch. It measures your heart rate and uses data to “talk” to your phone through the internet. This is an example of the “Internet of Things,” or a more familiar name called IoT. Now, let's apply this idea to cars. Cars can also “talk” to each other through the Internet, sharing important information like traffic conditions or road safety warnings. We call this technology the “Internet of Vehicles,” or IoV.

    These cars communicate by sending small messages known as data packets over the internet. Each packet carries useful information, but since all of them travel online, hackers can target them easily. Imagine you are driving and you receive false speed data sent by a hacker, potentially causing dangerous accidents. To keep everyone safe, we need smart systems that can quickly identify attacks and block them before actual damage happens. So nowadays, we use machine learning to teach computers to learn patterns from a bunch of data and spot attack behaviors.

    However, traditional machine learning methods require cars to send all their data to a central server for analysis. It's like everyone handing their detailed driving records directly to a single centralized office. This approach not only creates heavy traffic that slows down the communication but also raises privacy concerns. That's why I use a special approach called “Federated Learning.” Instead of sending everything to the central server, Federated Learning lets each car train its own small model locally. This means each local model learns what normal and attacked traffic messages look like inside the car, and only some of the summarized information is shared with the central server. Let's think of it like many students taking their own notes at home, each of them focuses on their unique understanding. Instead of handing all their detailed notes to a single teacher, they only share brief summaries or key highlights. The teacher then combines these summaries into broader lessons that benefit everyone. So back to our research, this combined knowledge from each car now becomes part of a “collective mind,” enabling cars to quickly and effectively recognize attacks without overwhelming the communication network while also protecting privacy.

    And here's the result: I've tested my system, and Federated Learning shows better accuracy than existing methods and with it, it's also harder for hackers to trick. Sooner or later, I hope we can step into our car with a peace of mind, knowing it's protected, and that our roads are safer for everyone.

Instructor's Commentary and Analysis

    The following analysis examines the success factors of this 3MT presentation from six perspectives, highlighting its strengths in communication strategy application. This student presentation effectively transforms complex technical concepts into accessible narratives, excelling in analogies, structure, and tone management — a valuable example for other contestants.

1. Start with the Familiar
    “Imagine you're wearing a smartwatch. It measures your heart rate and uses data to 'talk' to your phone through the Internet.”
    This script opens with an everyday image — a smartwatch — that immediately connects to the audience, making it an ideal entry point. It translates the abstract idea of the “Internet of Things” (IoT) into a real-world example, helping the audience grasp its function. In your presentation, it is a good idea to begin with something people already know. Think about daily technologies, shared experiences, or well-known situations that relate to your research. This sets a tone of invitation and inclusiveness, easing listeners into your subject.

2. Use Effective Analogies
    “It's like everyone handing their detailed driving records…”
    “Let's think of it like many students taking their own notes…”
    Analogies are the backbone of this script. Analogies bridge the gap between the known and the unknown by linking complex ideas to relatable experiences in a way that clarifies without losing the essence. The author's comparison of cars to students taking notes, for example, simplifies the concept of Federated Learning while preserving its core idea, makes it clearer and more memorable. So, develop analogies that suit your research. Think about how your research process mirrors something people already understand, like working on a team, planning a trip, assembling furniture, or troubleshooting a washing machine (well, this might be a little tricky to use, though not impossible). Find metaphors in your daily life. If your work involves algorithms, can you relate it to organizing a bookshelf? If it involves chemistry, maybe it's like mixing paint colors or baking?

3. Build a Well-structured Narrative
    The script follows a logical progression, moving from everyday technology, like smartwatches, to the Internet of Vehicles (IoV), then to emerging privacy concerns, the proposed solution of Federated Learning, its advantages, and finally, the promising results. When each new idea builds upon the last, the listener can easily follow the flow of thought without getting overwhelmed. Thus, lead your audience through a clear and logical storyline: starting with something familiar, introducing a rising threat, giving an innovative solution, and ending with a hopeful outlook. A well-structured narrative not only maintains engagement but also allows your listener to understand and remember your key points.

4. Deal with Jargons
    The script does not avoid jargon altogether; doing so would risk oversimplifying the research. Rather, it introduces technical terms in plain language, and only when necessary. For instance, the term “Internet of Things (IoT)” is introduced through a straightforward example: a smartwatch that monitors your heart rate and communicates with your phone. Only after establishing this image does the speaker name the actual term. Similarly, the “Internet of Vehicles (IoV)” is introduced by extending the same idea to cars, thus creating a seamless link. This strategy enables the audience to absorb new terms by first understanding how they relate to shared experiences.
    Other technical terms like “machine learning” and “Federated Learning” are also made accessible. Rather than listing definitions, the speaker shows how these technologies work using relatable metaphors, so the concepts are easy to visualize and digest. The audience can grasp what the speaker wants to convey without needing to comprehend the mathematics behind model training. When preparing for your 3MT presentation, choose two or three core terms (ideally, no more than three) essential for understanding your research; they each should serve a purpose in the story being told. Introduce them with concrete examples or analogies, and then define each in one sentence or two. Remember, the goal is to show how these terms function and why they matter within the framework of your research.

5. Adopt a Professional yet Approachable Tone
    Adopting a professional yet approachable tone means speaking in a way that demonstrates expertise and credibility while remaining relatable to your audience. In a 3MT, you're performing (in a way), not reciting. You want your audience to have trust in your expertise while also sensing that you're engaging with them, instead of speaking at them. You want to stay confident and well-informed (if not entirely knowledgeable) without coming across as overly formal or distant. Think about what talking to a group of intelligent, curious friends who are not experts in your field feels like, and bring that tone into your presentation. Use language that sounds natural and direct, as if you're having a thoughtful conversation with these friends.
    In the script, sentences like “Let's apply this idea to cars…” or “Imagine you are driving…” convey a tone and language that are both professional and approachable—professional in how they are intended to link back to technical research, and approachable in how they help the audience follow along with ease. In writing your own script, use rhetorical questions (“Why does this matter?”, “How to make this happen?”), transition phrases (“Now, let's move on to…”, “Here is where it gets interesting…”, “Think of it this way…”), and active verbs and familiar imagery. Repeat key words or ideas to reinforce important points, and insert pauses where appropriate to create a conversational flow. Most importantly, practice your script aloud over and over again. This allows you to hear how it actually sounds, identify awkward phrasing, and fine-tune your cadence. The more you practice aloud, the more confident and comfortable you will become.

6. Conclude with a Strong, Hopeful Message
    “Sooner or later, I hope we can step into our car with peace of mind, knowing it's protected, and that our roads are safer for everyone.”
    A closing remark like this shifts from research to vision. It connects the researcher's work to a broader goal: peace of mind, safety, and trust in technology. This helps the audience see its significance beyond the technical details. The presenter ends with a message that feels meaningful and optimistic. When crafting your conclusion, bring the focus back to your audience. Take a step back and show how your work fits into a bigger picture. Who gains from your results? What might change? The people you address, society, or your field? What future could this work help shape? Ending on a forward-looking note gives a sense of purpose and leaves your audience with a feeling of connection and optimism.

Putting It All Together
    The script “Your Car's Secret Bodyguard” succeeds because it transforms technical research into an intriguing narrative that resonates with a non-specialist audience. It balances expertise and simplicity by making sophisticated machine-learning concepts accessible without compromising depth. It demonstrates that even advanced science communication can be achieved clearly and memorably within three minutes. When preparing for your own 3MT presentation, think beyond theories, models, or figures. Prioritize clarity, flow, and relatability. Practice speaking with enthusiasm and sincerity. And most importantly, always keep your audience in mind — because at the heart of a great research talk is not (just) what you know, but how you share it.