Anan Yaghmour’s Post

This is the exact reason why most chips are usually soldered in.

Boiling (hard) water reduces #microplastics! As a Dutch born Chinese, I always got questions from many Dutch friends why I am drinking hot water (without a teabag in it). Drinking warm water that has been boiled and cooled down to a drinkable temperature is common practice in Asian countries such as China and Japan. Clear advantages of boiling water: it kills pathogens and warms up the body. Traditional Chinese medicin also believes in balance: the body is warm, so the water we intake should also be warm. Now another adventage: reduction of microplastics. A recent paper has showed a reduction of up to 90% of three types of microplastics from water that has been: 1. Boiled 2. Cooled down to room temperature 3. Filtered over a coffee filter The idea is that calcium carbonate (also known as limestone or chalk) crystallises and entraps the microplastics. These particles can then we filtered out after cooling. Simple but effective. Just a side note that this works better for #hardwater, which have higher calcium carbonate contents. Now I have an additional explanation when someone asks me why I always boil my water. Am curious to see more #research on this topic and if this would lead to any (inter)national health advices in the near future!

I'm on a spiritual journey to refine my understanding of generative models especially diffusions and flows. Both follow a simple principle, continuously change data until they match target data points then generate new data using the transformed distribution. The difficulty is of course finding the transformed distribution. The ultimate equation that governs such generative models and also is the roadmap for discovering other variants was proposed by Kolmogorov in his 1931 paper: "Über die analytischen Methoden in der Wahrscheinlichkeitsrechnung" Later in 1949, William Feller proposed a more general form and called it Kolmogorov's forward equation. This equation is perhaps the most general of its kind as we can derive other similar equations like Fokker-Planck or Kramers–Moyal expansion as special cases. Image: An Introduction to Probability Theory, Feller, Vol II, Chapter X

ماعلاقة الإحصاء بتعلم الآلة؟ كيف لي فهم تعلم الآلة من دون معرفة الاحتمالات و تنبؤات، كيف لي أن أتنبأ بقيمة معينة باستخدام التعلم الآلة؟ لنعلم أن بوابة فهم تعلم الآلة هو معرفة أساسيات الإحصاء، لذلك لنعرف أن الإحصاء له علاقة وثيقة بتعلم الآلة وتحليل البيانات. إليك بعض النقاط التي توضح العلاقة بين الإحصاء وتعلم الآلة: ١-الاعتماد على النماذج الإحصائية: اغلب نماذج التعلم الآلة معتمدة على نماذج الاحصائية سبيل المثال، يمكن استخدام الانحدار الخطي والتحليل العاملي والتحليل التجميعي وغيرها من النماذج الإحصائية لفهم العلاقات والأنماط في البيانات. ٢- الاستنتاج الإحصائي: يتم استخدام الاحتمالات وهي احد أساسيات الإحصاء،. ٣- تقليل الأبعاد: يعالج الإحصاء قضية تقليل الأبعاد، وهي عملية تقليل الأبعاد من مجموعة كبيرة من المتغيرات إلى مجموعة أصغر تحتوي على المعلومات الأساسية. وهي أحد أساليب Data mining وهي أساليب تعلم الآلة. ٤- التحقق من النماذج: يستخدم الإحصاء في التحقق من يمكن استخدام اختبارات الفرضية وتحليل الانحدار وتقنيات التحقق الأخرى للتأكد من أن النماذج تعكس البيانات بشكل دقيق ويمكنها التعميم على بيانات جديدة. هذه أهم النقاط من وجهة نظري مرفق لكم ملخص جميل عن أهم أساسيات الإحصاء. #الذكاء_الاصطناعي #تعلم_الآلة

“Data centric” AI is a dangerous viral concept that is currently sweeping through the AI community, for which I hope I can provide a vaccine. Data has no value or meaning in the abstract, without an understanding of the context or processes by which it was generated. This is the same reason that the term “data science” is an oxymoron - there is no “science” of data in the abstract. Only of specific physical processes that generate data. The only way that a 100% focus on data could work is if the data had such complete and detailed coverage of the entire space of possibilities that the outcome would basically be predetermined, and could be obtained by any interpolation method you choose. But that’s neither feasible nor economically viable. Even LLMs show that this “data centric” view is wrong. It wasn’t until the introduction of a special functional form, the transformer, in 2017, that incorporated the domain-specific knowledge that the structure of language relies on context-dependent long range associations, that truly plausible dialog could be modeled. Previous, generic interpolation methods could not have achieved such results even using the entire internet as training data. #data #artificialintelligence [Comment added 12/16: machine learning is the subset of modeling that requires data and cannot be done in closed form. So of course data is important. But the “data centric” slogan is dangerous because it deliberately takes people’s attention away from the rest of the modeling problem, which is already overly neglected.]

⚒Write a PowerPoint Presentation with ChatGPT🚀👉 https://lnkd.in/gM65CNAr - ChatGPT Mastery - Prompt 👇 👇👇 Prompt: "I want you to write me VBA code for a PowerPoint presentation about the history of AI. You are to fill in all the text with your own knowledge, no placeholders. I need 5 slides." #chatgpt #ai #chatgpt4 #openai