<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by ross. on Medium]]></title>
        <description><![CDATA[Stories by ross. on Medium]]></description>
        <link>https://medium.com/@rossvell?source=rss-f9c3325ae8c1------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*lfktXS9W3yIeZcU8rouq-w@2x.jpeg</url>
            <title>Stories by ross. on Medium</title>
            <link>https://medium.com/@rossvell?source=rss-f9c3325ae8c1------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 13:22:38 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@rossvell/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Particle Identification in High-Energy Physics Using Machine Learning]]></title>
            <link>https://medium.com/@rossvell/particle-identification-in-high-energy-physics-using-machine-learning-bc9eff954f56?source=rss-f9c3325ae8c1------2</link>
            <guid isPermaLink="false">https://medium.com/p/bc9eff954f56</guid>
            <dc:creator><![CDATA[ross.]]></dc:creator>
            <pubDate>Wed, 20 May 2026 07:12:55 GMT</pubDate>
            <atom:updated>2026-05-20T07:16:26.335Z</atom:updated>
            <content:encoded><![CDATA[<h3>Final Assignment Plan for Research Based Learning (RBL) in FI3271 Data Analysis with Machine Learning</h3><h3>Introduction</h3><p>Particle physics is one of the most fundamental branches of physics that studies the smallest constituents of matter and the interactions between them. Modern particle physics experiments, such as those conducted at the European Organization for Nuclear Research (CERN), generate enormous amounts of detector data from high-energy particle collisions. One of the major challenges in these experiments is identifying what type of particle produced a certain detector signal.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*CQexZNuvD-cbMRJm" /><figcaption>Particle Collision in the CMS Experiment at the LHC CERN</figcaption></figure><p>Traditionally, particle identification relies on detector physics and statistical analysis. However, with the rapid development of artificial intelligence and machine learning, researchers are now able to classify particle events more efficiently using computational methods.</p><p>This project aims to explore the application of machine learning techniques for particle identification using high-energy physics datasets. The project is designed as a Research Based Learning (RBL) assignment and focuses on building a simple but meaningful machine learning pipeline suitable for undergraduate students.</p><h3>Tentative Title</h3><p><strong>Particle Identification in High-Energy Physics Using Machine Learning</strong></p><p>Alternative titles:</p><ul><li>Machine Learning-Based Particle Classification Using Detector Data</li><li>Classification of Particle Collision Events Using Python</li><li>Application of Machine Learning in High-Energy Physics</li></ul><h3>Problem to Solve</h3><p>High-energy particle detectors record a large number of variables from collision events, such as momentum, energy, and detector responses. The challenge is determining what type of particle produced those signals.</p><p>In particle physics experiments, correct particle identification is important because it helps researchers:</p><ul><li>distinguish signal events from background noise,</li><li>study fundamental interactions,</li><li>analyze rare particle decays,</li><li>and search for new physics phenomena.</li></ul><p>The main research question in this project is:</p><blockquote><em>“How accurately can machine learning classify particle collision events based on detector observables?”</em></blockquote><p>This project will attempt to solve the classification problem using machine learning algorithms trained on publicly available particle physics datasets.</p><h3>Programming Language, Libraries, and Platform</h3><h3>Programming Language</h3><ul><li>Python</li></ul><h3>Libraries</h3><p>The following libraries will be used:</p><ul><li>NumPy → numerical computation</li><li>Pandas → data processing and analysis</li><li>Matplotlib → visualization</li><li>Scikit-learn → machine learning algorithms and evaluation</li></ul><p>Optional libraries:</p><ul><li>Seaborn → advanced visualization</li><li>XGBoost → improved classification performance</li></ul><h3>Platform</h3><ul><li>Google Colab</li><li>Jupyter Notebook</li></ul><p>These platforms are chosen because they are free, accessible, and suitable for machine learning development.</p><h3>Dataset Candidate</h3><p>The project will use publicly available particle physics datasets.</p><h3>Main Dataset Candidate</h3><h3>Higgs Boson Machine Learning Challenge Dataset</h3><p><a href="https://www.kaggle.com/competitions/higgs-boson/data">https://www.kaggle.com/competitions/higgs-boson/data</a></p><p>This dataset contains simulated particle collision events and is widely used for machine learning studies in high-energy physics.</p><p>Possible features include:</p><ul><li>momentum,</li><li>energy-related variables,</li><li>angular distributions,</li><li>detector observables.</li></ul><p>The dataset labels indicate whether an event is likely a signal event or background event.</p><h3>Additional Dataset References</h3><h3>CERN Open Data Portal</h3><p><a href="https://opendata.cern.ch/">https://opendata.cern.ch/</a></p><h3>Kaggle Particle Physics Datasets</h3><p><a href="https://www.kaggle.com/datasets?search=particle+physics">https://www.kaggle.com/datasets?search=particle+physics</a></p><h3>Methodology</h3><p>The general workflow of the project is:</p><ol><li>Literature study on particle identification and machine learning</li><li>Dataset collection and preprocessing</li><li>Exploratory Data Analysis (EDA)</li><li>Feature selection and normalization</li><li>Training machine learning models</li><li>Model evaluation and comparison</li><li>Analysis and interpretation of results</li><li>Final report and presentation</li></ol><h3>Machine Learning Models</h3><p>Several machine learning classification algorithms will be tested and compared.</p><h3>Candidate Models</h3><ul><li>Decision Tree</li><li>Random Forest</li><li>K-Nearest Neighbor (KNN)</li></ul><p>Optional advanced model:</p><ul><li>XGBoost</li></ul><p>The performance of each model will be evaluated using:</p><ul><li>Accuracy</li><li>Precision</li><li>Recall</li><li>F1-score</li><li>Confusion Matrix</li></ul><h3>Timeline of Work</h3><p>The project is planned to be completed within approximately two weeks.</p><p>Week 1: Literature study, dataset exploration, preprocessing, and baseline model training</p><p>Week 2: Model optimization, evaluation, visualization, analysis, report writing, and presentation preparation</p><h3>Estimation Number of Team Members</h3><p>A team of 2–3 students is considered ideal because the project scope is moderate and manageable within the given timeline.</p><h3>Mock-Up of Expected Output</h3><p>The expected outputs of the project include:</p><ul><li>Machine learning classification results</li><li>Confusion matrix visualization</li><li>Accuracy comparison graphs</li><li>Feature importance plots</li><li>Research-style report</li><li>Final presentation slides</li></ul><p>Possible visualization examples:</p><h3>Confusion Matrix</h3><p>A heatmap showing the classification performance of the machine learning model.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/595/1*MPHmJp4064UsmeAd9AU8Zw.png" /><figcaption><a href="https://chatgpt.com/s/t_6a0d5da4321c8191ba701330f59e486d">https://chatgpt.com/s/t_6a0d5da4321c8191ba701330f59e486d</a></figcaption></figure><h3>Feature Importance Plot</h3><p>A graph showing which detector variables contribute the most to classification.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/813/1*rwCc8AKDrr60tXsB8eJqzw.png" /><figcaption><a href="https://chatgpt.com/s/t_6a0d5dff2d8c8191858c0723848d7947">https://chatgpt.com/s/t_6a0d5dff2d8c8191858c0723848d7947</a></figcaption></figure><h3>Accuracy Comparison</h3><p>A comparison chart between different machine learning algorithms.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/686/1*t65uUZW_KJ3Fau9DehaRpA.png" /><figcaption><a href="https://chatgpt.com/s/t_6a0d5e77db408191965b0fe0395b555b">https://chatgpt.com/s/t_6a0d5e77db408191965b0fe0395b555b</a></figcaption></figure><h3>Expected Outcomes</h3><p>At the end of the project, the team is expected to:</p><ul><li>understand the basics of particle identification,</li><li>learn the workflow of machine learning in physics,</li><li>gain experience in scientific data analysis,</li><li>evaluate classification performance,</li><li>and interpret computational results within the context of high-energy physics.</li></ul><p>This project is expected to demonstrate how computational techniques can support modern particle physics research.</p><h3>Conclusion</h3><p>Machine learning has become an important tool in modern high-energy physics research. Through this project, students can explore the application of machine learning techniques for particle identification using real-world datasets inspired by particle collider experiments.</p><p>The project is suitable for undergraduate students because it combines physics concepts, data science, and programming within a manageable scope and timeline. In addition, the project provides hands-on experience with research-oriented workflows commonly used in contemporary scientific studies.</p><h3>Link to Generative AI</h3><p>This medium story was created with the help of ChatGPT.</p><p>Link: <a href="https://chatgpt.com/share/6a0d5ed6-6c70-83ec-a0eb-8e6028007c10">https://chatgpt.com/share/6a0d5ed6-6c70-83ec-a0eb-8e6028007c10</a></p><h3>References</h3><p>[1] Baldi, P., Sadowski, P. &amp; Whiteson, D., Searching for Exotic Particles in High-Energy Physics with Deep Learning, Nature Communications, 5, pp. 4308, 2014. (Journal)</p><p>[2] Roe, B. P., Yang, H. J., Zhu, J., Liu, Y., Stancu, I. &amp; McGregor, G., Boosted Decision Trees as an Alternative to Artificial Neural Networks for Particle Identification, Nuclear Instruments and Methods in Physics Research A, 543(2–3), pp. 577–584, 2005. (Journal)</p><p>[3] Guest, D., Cranmer, K. &amp; Whiteson, D., Deep Learning and Its Application to LHC Physics, Annual Review of Nuclear and Particle Science, 68, pp. 161–181, 2018. (Journal)</p><p>[4] CERN Open Data Portal, CERN, <a href="https://opendata.cern.ch/">https://opendata.cern.ch/</a>, (20 May 2026). (Full URL Link)</p><p>[5] Higgs Boson Machine Learning Challenge Dataset, Kaggle, <a href="https://www.kaggle.com/competitions/higgs-boson/data">https://www.kaggle.com/competitions/higgs-boson/data</a>, (20 May 2026). (Full URL Link)</p><p>[6] Pedregosa, F. et al., Scikit-learn: Machine Learning in Python, Journal of Machine Learning Research, 12, pp. 2825–2830, 2011. (Journal)</p><p>[7] Chollet, F., Deep Learning with Python, 2nd ed., Manning Publications, 2021. (Book)</p><p>[8] Bishop, C. M., Pattern Recognition and Machine Learning, Springer, 2006. (Book)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bc9eff954f56" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Comparing 5 Generative AIs Through Infographic Design]]></title>
            <link>https://medium.com/@rossvell/comparing-5-generative-ais-through-infographic-design-92f5a22f2010?source=rss-f9c3325ae8c1------2</link>
            <guid isPermaLink="false">https://medium.com/p/92f5a22f2010</guid>
            <dc:creator><![CDATA[ross.]]></dc:creator>
            <pubDate>Mon, 27 Apr 2026 06:52:51 GMT</pubDate>
            <atom:updated>2026-04-27T06:52:51.988Z</atom:updated>
            <content:encoded><![CDATA[<p><em>How different AI models interpret the same prompt, and what we can learn from it.</em></p><h3>Introduction: Why Compare Generative AI?</h3><p>Artificial Intelligence is no longer something futuristic, it’s already part of our daily lives. From recommendation systems to face recognition, many of the tools we use today rely on <strong>neural networks</strong> working behind the scenes.</p><p>At the same time, generative AI tools like ChatGPT, Gemini, Claude, Copilot, and Perplexity are becoming increasingly powerful. But a simple question arises:</p><p><em>“Do different AI models produce the same results when given the exact same prompt?”</em></p><p>To explore this, I conducted a small experiment. I gave <strong>five different generative AI assistants</strong> the same prompt and asked them to create an infographic about neural networks in daily life.</p><p>The goal was not just to see the outputs but to understand:</p><ul><li>How each AI interprets instructions</li><li>How they balance creativity vs constraints</li><li>Where they succeed — and where they make mistakes</li></ul><h3>The Prompt Used</h3><p>To ensure fairness, I used <strong>one identical one-shot prompt</strong> for all five AIs:</p><blockquote>Create an infographic about “Neural Networks in Daily Life”.</blockquote><blockquote>Constraints:<br>1. The infographic must include:<br> — A short definition of neural networks (max 30 words)<br> — At least 5 real-life applications (e.g., recommendation systems, face recognition, medical diagnosis, etc.)<br> — A simple explanation of how neural networks work (non-technical, max 50 words)<br> — One section on advantages and one on limitations (each max 3 bullet points)</blockquote><blockquote>2. Design requirements:<br> — Use a clean, modern layout<br> — Include icons or visual elements for each application<br> — Use no more than 3 main colors<br> — Text must be concise and readable</blockquote><blockquote>3. Output format:<br> — Generate a complete infographic (image or structured layout if image is not supported)<br> — If image generation is not available, provide a detailed visual layout description</blockquote><blockquote>4. Tone:<br> — Educational and beginner-friendly</blockquote><blockquote>Important:<br>Do not exceed the word limits. Keep it visually structured and easy to scan.</blockquote><p>This prompt was designed to test both content understanding and visual/structural organization.</p><h3>Results from Each AI</h3><ol><li><strong>ChatGPT:</strong></li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/695/1*HWUCCIplxTTYeULq-3zmhg.png" /><figcaption><a href="https://chatgpt.com/share/69ef014b-67c0-8320-baca-39fcce6b1108">ChatGPT-prompt</a></figcaption></figure><p>ChatGPT’s result felt structured and clear from the start. The sections were well-organized, and the explanations were easy to follow. It leaned toward clarity rather than visual creativity, almost like a well-designed study note rather than a highly visual infographic. Still, it respected most constraints quite well.</p><p><strong>2. Gemini:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4FG3JCSecvjrNVjo_4GgPg.png" /><figcaption><a href="https://gemini.google.com/share/c492efe0d7b0">Gemini-prompt</a></figcaption></figure><p>Gemini’s output felt more “visual” in its thinking. The layout descriptions were smoother and more design-oriented. However, in some parts, it simplified the content a bit too much and occasionally stretched beyond the word limits. It felt intuitive, but slightly less strict.</p><p><strong>3. Copilot:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_ZvpH94lfdcgBAABY2uomw.png" /><figcaption><a href="https://copilot.microsoft.com/shares/h694bPF35dLqa5FXWqbzV">Copilot-prompt</a></figcaption></figure><p>Copilot delivered something concise and efficient. It followed the structure quite closely and didn’t overcomplicate things. However, compared to others, it felt a bit more rigid and less creative in presentation. It worked, but it didn’t stand out visually.</p><p><strong>4. Claude:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/491/1*nSSCSVD3IGwQ3g0yimCTbA.png" /><figcaption><a href="https://claude.ai/share/458d53ae-f37f-4c42-9ecc-ba47806f3b05">Claude-prompt</a></figcaption></figure><p>Claude’s strength was immediately noticeable in its explanations. Everything felt very beginner-friendly and thoughtfully written. However, it tended to be slightly more verbose than required, and the “infographic feel” wasn’t always as strong, it leaned toward explanation rather than design.</p><p><strong>5. Perplexity</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*nuIIMZXOA58aXK0TVWn7ug.png" /><figcaption><a href="https://www.perplexity.ai/search/e444431a-13a8-4b77-a170-e0500703d434">Perplexity-prompt</a></figcaption></figure><p>Perplexity focused heavily on delivering accurate and informative content. The explanations were solid, but the output sometimes felt more like a structured summary than an infographic. It prioritized clarity of information over visual storytelling.</p><h3>What Actually Changed Between the AIs?</h3><p>At first, I expected small differences.</p><p>But the more I looked at the results, the more I realized — each AI wasn’t just answering differently.</p><p>They were <strong>thinking differently</strong>.</p><p>ChatGPT approached the task like a structured communicator. It carefully followed the instructions and produced something balanced, though slightly text-heavy. Gemini, on the other hand, seemed to “think visually” first, it cared more about how the infographic would feel, even if it meant being a bit loose with constraints.</p><p>Copilot felt the most disciplined. It followed the rules closely, stayed concise, and delivered exactly what was asked, but without much personality. Claude stood out in a different way. It didn’t just answer, it explained. You could feel the intention to teach, even if that sometimes made it longer than necessary.</p><p>Perplexity took yet another direction. It behaved almost like a researcher focused on accuracy, clarity, and information. But in doing so, it sometimes forgot that an infographic is not just about information, it’s about how that information is presented visually.</p><p>What surprised me most was this:<br>Even with the exact same prompt, the outputs weren’t just variations, they reflected completely different priorities.</p><p>Some AIs optimized for <strong>clarity</strong>, others for <strong>design</strong>, others for <strong>accuracy</strong>, and some for <strong>instruction following</strong>.</p><p>And that’s where the real insight lies.</p><h3>So… Which One Is Better?</h3><p>After comparing all five results, I would choose <strong>ChatGPT</strong> as the most effective overall. While other models show specific strengths, such as Gemini’s visual intuition or Claude’s improved design, ChatGPT provides the best balance between following instructions, clarity of explanation, and structured presentation. It consistently meets the given constraints, delivers content that is easy to understand for beginners, and organizes information in a way that fits the purpose of an infographic. Although it may not be the most visually creative, its reliability and clarity make it the strongest choice for this task.</p><h3>Final Thoughts</h3><p>This small experiment changed how I see generative AI.</p><p>Even when given the exact same prompt, each model interprets the task differently. Some prioritize clarity, others design, others accuracy.</p><p>And maybe that’s the most important takeaway:</p><p>AI doesn’t just generate answers, it reflects its own way of “thinking.”</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=92f5a22f2010" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From “I’ll Never Code Again” to Machine Learning in Particle Physics]]></title>
            <link>https://medium.com/@rossvell/from-ill-never-code-again-to-machine-learning-in-particle-physics-1f2fba4b40fc?source=rss-f9c3325ae8c1------2</link>
            <guid isPermaLink="false">https://medium.com/p/1f2fba4b40fc</guid>
            <dc:creator><![CDATA[ross.]]></dc:creator>
            <pubDate>Wed, 22 Apr 2026 09:58:22 GMT</pubDate>
            <atom:updated>2026-04-22T22:05:14.866Z</atom:updated>
            <content:encoded><![CDATA[<p>Three years ago, I made a quiet promise to myself: <em>I would never code again.</em></p><p>As a first-year student at Institut Teknologi Bandung, I was introduced to computation through a mandatory course called <em>Introduction to Computation</em>. At that time, everything felt new and honestly, a bit intimidating. I remember questioning myself a lot: <em>Can I really do this? Am I thinking the right way?</em></p><p>But as the weeks went by, something slowly shifted.</p><p>I started to enjoy the process. Not necessarily the coding itself at first, but the way it trained my mind. It taught me how to think step by step, how to organize ideas, and how to approach problems more systematically. There was something satisfying about turning something complicated into something structured and solvable.</p><p>Even so, when the course ended, my grade didn’t turn out the way I had hoped. It wasn’t a failure, but it left me feeling a bit discouraged. Not because I hated the subject, but more because I felt like I hadn’t fully reached my potential there. So, without making a big decision about it, I just… drifted away from coding.</p><p>At that point in my academic journey, I was very sure about one thing: I wanted to become a theoretical physicist.</p><p>So I focused my energy there. I chose courses that aligned with that goal and didn’t really explore areas like data analysis or programming. It wasn’t that I rejected them, it just didn’t feel relevant to the path I had in mind at the time.</p><p>For a while, everything felt aligned.</p><p>Until I was introduced to something that changed the direction of my interest entirely.</p><h3>Chasing a Dream in High Energy Physics</h3><p>I was introduced to high energy physics (particle physics), and I immediately felt a strong connection to it. There was something about it that felt alive, dynamic, and deeply intriguing.</p><p>But along with that excitement came a realization: if I wanted to pursue this field, especially from an experimental perspective, I couldn’t avoid computation anymore.</p><p>In fact, it’s deeply embedded in the work.</p><p>To understand this better, we can look at places like CERN, where experiments in particle physics are conducted on a massive scale. The detectors there record an enormous amount of data from particle collisions, so large that it goes far beyond what we usually imagine when we think about “big data.”</p><p>And this is where the role of coding becomes unavoidable.</p><p>As experimental physicists, we don’t just observe results, we process raw data, clean it, analyze it, simulate physical processes, and try to extract meaningful patterns from it. Almost everything we do involves computational tools.</p><p>Over time, I began to understand that coding in this context is not just a technical skill, it’s a language. A way to communicate with data.</p><h3>Where Machine Learning Comes In</h3><p>As the amount of data keeps growing, traditional analysis methods are often not enough. This is where machine learning becomes incredibly valuable.</p><p>In high energy physics, machine learning helps in many ways:</p><ul><li>distinguishing between signal and background events</li><li>identifying rare particle interactions</li><li>improving the efficiency of data analysis</li><li>assisting in pattern recognition that might be too complex for manual methods</li></ul><p>It doesn’t replace physics intuition, but it enhances it.</p><p>And that’s when things started to click for me.</p><h3>Coming Back, But With a Different Perspective</h3><p>This realization is what led me to take <strong>FI3271 Data Analysis with Machine Learning</strong>.</p><p>If three years ago I saw coding as a challenging course I had to pass, now I see it as a tool I <em>need</em> to understand the field I care about.</p><p>Still, I won’t pretend it was easy to come back.</p><p>I felt nervous at the beginning. It had been a long time since I last wrote code, and I had never taken a course specifically focused on data analysis before. There were moments where I felt a bit behind compared to others.</p><p>But instead of stepping back, I decided to stay curious.</p><p>I try to follow the lectures carefully, revisit the materials, and do my best in every assignment. I may not understand everything instantly, but I’m learning to be patient with the process.</p><h3>What I’ve Learned So Far</h3><p>Looking back at the first half of this course, I realize that I’ve been introduced to concepts that once felt distant, but are now starting to make sense.</p><p>We began with the fundamentals: understanding the relationship between Artificial Intelligence, Machine Learning, and Deep Learning. I learned that machine learning itself can be divided into supervised and unsupervised learning, depending on whether the data comes with labels or not. This helped me see that not all problems are approached in the same way, and choosing the right framework matters.</p><p>Then we explored <strong>Principal Component Analysis (PCA)</strong>, which opened my eyes to how high dimensional data can be simplified without losing its essential structure. This felt especially relevant to physics, where datasets can quickly become complex.</p><p>We also studied <strong>Support Vector Machine (SVM)</strong>, where I learned about finding optimal boundaries that separate data into classes. What stood out to me was not just the classification itself, but the idea of maximizing the margin finding solutions that are not only correct, but also stable.</p><p>And finally, <strong>k-Nearest Neighbors (kNN)</strong> showed me that sometimes, simple approaches can still be powerful. By looking at nearby data points, we can make meaningful predictions without overly complicated models.</p><p>Each of these topics gave me a small but important insight: understanding the data is just as important as choosing the method.</p><h3>What I Hope to Gain</h3><p>By the end of this course, I hope to build a foundation that I can carry into my future research.</p><p>I want to be able to look at a dataset and not feel lost. I want to understand what tools I can use, how to approach a problem, and how to interpret the results meaningfully. Especially in the context of high energy physics, where data plays such a central role, these skills feel essential.</p><p>Even more than that, I want to regain confidence in working with computation not perfectly, but comfortably.</p><h3>A Different Kind of Final Exam</h3><p>If I could choose, I would prefer a project-based final exam, ideally done in a group.</p><p>Because learning like this feels more real.</p><p>Working on a project allows us to connect concepts, explore ideas more deeply, and face challenges that don’t have a single correct answer. Doing it in a team also reflects how research actually works. We exchange ideas, support each other, and learn collaboratively.</p><p>Compared to a written exam, a project feels more like a learning experience than just an evaluation.</p><h3>Closing: Why This Course Matters to Me</h3><p>Taking this course feels like coming back to something I once left behind, but with a new understanding.</p><p>This time, I’m not learning computation just to pass a class. I’m learning it because I see where it can take me.</p><p>And even if, somewhere along the way, I don’t end up pursuing particle physics, I know this experience will still matter.</p><p>Because the ability to think analytically, to work with data, and to approach problems systematically, those are skills that go beyond any single field.</p><p>In a way, this course is not just about machine learning.</p><p>It’s about growth, direction, and giving myself another chance to learn something I once thought I couldn’t do.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1f2fba4b40fc" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>