Day 78: ✅ Focused on experimental design, evaluation metrics, and reproducibility setup for upcoming TPG–FPGA experiments ✅ Continued simulation runs and further collected output data for my new paper 📊 ✅ 60 minutes of workout and dance 🏋🏻‍♀️💃🏻 #ML #AI #FPGA #100DaysofCode

Day 77: ✅ Focused on timing closure and resource balance in TPG pipelines to keep II = 1 at 250 MHz ✅ Obtained partial simulation results for my new paper 📊 ✅ 45 minutes workout 🏋🏻‍♀️ #MachineLearning #AI #FPGA #100DaysofCode

A quiet Sunday walk! Sometimes nature debugs the mind better than any code 🌿🧠

Day 76: ✅Explored how TPGs map onto FPGA hardware, from LUTs and DSPs to pipelining and parallel execution, and why FPGA’s spatial dataflow is the perfect match for gradient-free TPG learning ✅45-minute workout and dance🏋🏻‍♀️💃🏻 #MachineLearning #AI #FPGA #100DaysofCode

Day 75: ✅ Explored the core structure of TPG and how they evolve their architecture through adaptation instead of gradient-based training👩🏻‍💻✍️🏻 ✅40-minute workout🤸🏻‍♀️ #MachineLearning #AI #100DaysofCode

Day 74: ✅ Explored the foundation of gradient-free learning in Tangled Program Graphs ✅ 60-minute walk and dance💃🏻🔋 #100DaysofMachineLearning #AI

Day 73: ✅Reviewed the core Machine Learning principles forming the basis of the TPG–FPGA project ✅Started drafting a new conference paper ✅40-minute workout #MachineLearningProjects

Back after a long break 🚀 This time, I’m diving into something I truly love: Tangled Program Graphs (TPG) on FPGA 💻🧠 I’ll share a tiny spark every night. One concept, one idea, one step closer! ⚡️⚙️ #MachineLearning #TPG #ResearchJourney

which struggle with remembering information over many time steps due to vanishing gradients, LSTMs can selectively remember or forget information, making them effective for tasks like time series prediction, speech recognition, and language modeling. 2/ #100DaysOfCode #Python

Day 72: ✅LSTM (Long Short-Term Memory) 📌PS: LSTM is a special type of RNN designed to better capture long-term dependencies in sequential data by using memory cells and gates that regulate the flow of information. Unlike traditional RNNs, 🧵1/

Which is addressed by advanced models like LSTMs and GRUs. 3/ #100DaysOfCode #Python #MachineLearning

This allows them to learn patterns over time. They're useful in tasks like language modeling, time series prediction, and speech recognition. However, RNNs can struggle with long-term dependencies due to the vanishing gradient problem, 2/

Day 71: ✅RNN (Recurrent Neural Network) 📌PS: RNN is a neural network designed for sequential data. Unlike traditional networks, RNNs use feedback loops, where the output from the previous time step is used as input for the next. 🧵1/

She touched infinity and left fingerprints! #MaryamMirzakhani

As the depth increases, the network can extract higher-level information, making DCNNs very powerful for tasks like image classification, object detection, and face recognition. 3/ #100DaysOfCode #Python #MachineLearning

It uses convolutional layers to automatically learn spatial features by sliding filters over the input. CNNs are effective at capturing patterns like edges, textures, and shapes. A DCNN is simply a CNN with many layers, allowing it to learn more complex and abstract features. 2/

Day 70: ✅ DCNN (Deep Convolutional Neural Network) 📌PS: A CNN is a type of deep learning model specially designed for processing data with a grid-like structure, such as images. 🧵1/

Additionally, it can show which features are most important, making it popular in fields like medical diagnosis, fraud detection, and recommendation systems. 3/ #100DaysOfCode #python #MachineLearning

and the final result comes from majority voting (for classification) or averaging (for regression). This approach improves accuracy and reduces overfitting compared to a single decision tree. It also handles missing data well and works efficiently with large datasets. 2/

Day 69: ✅Random Forest 📌PS: Random Forest is an ensemble machine learning algorithm that builds multiple decision trees using random subsets of data and features. Each tree makes a prediction, 🧵1/