Now I'm in Part 1: Introduction to Computer Vision (Progress 26%); Lesson 4: Image Representation & Classification; 19. Day and Night Classification
Day 3: September 26, 2019
I watched many video lectures and programmed functions in many Jupyter notebooks.
Now I'm in Part 1: Introduction to Computer Vision (Progress 46%); Lesson 5: Convolutional Filters and Edge Detection (Progress 73%); 19. Quiz: Hough Space
I encourage @KhushkumarP, @Jacqueline SusanM.
Day 4: September 27, 2019
I filled the form of the #30DaysofUdacity pledge.
I watched many video lectures and programmed functions in many Jupyter notebooks.
Now I'm in Part 1: Introduction to Computer Vision (Progress 81%); Lesson 8: CNN Layers and Feature Visualization (Progress 76%); 22. Visualizing CNNs (Part 2)
I encourage @KhushkumarP, @Jacqueline SusanM, @labibaR, @OudarjyaS.
Day 5: September 28, 2019
I watched many video lectures and programmed functions in many Jupyter notebooks.
Now I'm in Part 1: Introduction to Computer Vision (Progress 88%); Project: Facial Keypoint Detection; 3. Project: Facial Keypoint Detection
I'm working on Project 1: Facial Keypoint Detection. So far, my average loss is 0.0011, which is good.
I encourage @KhushkumarP, @Jacqueline SusanM, @labibaR, @OudarjyaS, @MuhammadN, @Mohamed ChoukriB.
Day 6: September 29, 2019
I'm about to submit the Project 1: Facial Keypoint Detection. So far, the results are good. See the attached photo.
I'm in Part 1: Introduction to Computer Vision (Progress 88%); Project: Facial Keypoint Detection; 3. Project: Facial Keypoint Detection
I had a virtual meetup with the people of the Deep Learning nanodegree.
I had a virtual meetup with the people of the Computer Vision nanodegree.
I'm in Part 3: Advanced Computer Vision & Deep Learning (Progress 91%); Project: Image Captioning
I'm programming Project 2: Image Captioning. However, I'm stuck because the COCO dataset is tricky to install. Perhaps I should work in the Udacity website.
I'm in Part 3: Advanced Computer Vision & Deep Learning (Progress 91%); Project: Image Captioning
I'm programming Project 2: Image Captioning. I finally understood the CNN Encoder and the RNN Decoder. I already programmed the RNN Decoder and I selected good-enough hyperparameters to train my neural networks. The learning curve is slowly and steadily converging, almost monotonically with minor jumps in the loss. See the photo. But it takes ages to train in the Udacity's GPU. I hope my 50 hours of GPU won't be consumed too fast.
I'm in Part 4: Object Tracking and Localization (Progress 67%); Lesson 6 - Matrices and Transformation of State (Progress 74%); 23. Matrix Multiplication
I finish watching all the video lectures of the core curriculum:
Part 1: Introduction to Computer Vision (Progress 100%);
Part 2: Optional - Cloud Computing (Progress 100%);
Part 3: Advanced Computer Vision & Deep Learning (Progress 100%);
Part 4: Object Tracking and Localization (Progress 100%).
I'm working in the final project: Project 3 - Landmark Detection & Robot Tracking (SLAM)
I encourage @KhushkumarP, @Jacqueline SusanM, @labibaR, @OudarjyaS, @MuhammadN, @Mohamed ChoukriB, @KonstantinosK, @AdityaK, @MahmmoudM, @LauraT, @ElenaK, @HelenaB, @RobO, @ShafaqA, @Nancy A.
Day 18: October 11, 2019
I submitted and passed the Project 3 - Landmark Detection & Robot Tracking (SLAM).
I graduated from the Computer Vision nanodegree.
I encourage @KhushkumarP, @Jacqueline SusanM, @labibaR, @OudarjyaS, @MuhammadN, @Mohamed ChoukriB, @KonstantinosK, @AdityaK, @MahmmoudM, @LauraT, @ElenaK, @HelenaB, @RobO, @ShafaqA, @Nancy A.
Day 19: October 12, 2019
I attended the virtual meetup of Deep Learning.
I attended the virtual meetup of Computer Vision.
I applied to the role of technical moderator of the Slack channel #cvnd_techhelp.
I encourage @KhushkumarP, @Jacqueline SusanM, @labibaR, @OudarjyaS, @MuhammadN, @Mohamed ChoukriB, @KonstantinosK, @AdityaK, @MahmmoudM, @LauraT, @ElenaK, @HelenaB, @RobO, @ShafaqA, @Nancy A.
Day 20: October 13, 2019
I'm watching all the video lectures of the extracurricular material.
I'm in Part 2. Review: Training A Neural Network (Progress 10%); Lesson 1 - Feedforward and Backpropagation (Progress 60%); 3. Neural Network Architecture.
I encourage @KhushkumarP, @Jacqueline SusanM, @labibaR, @OudarjyaS, @MuhammadN, @Mohamed ChoukriB, @KonstantinosK, @AdityaK, @MahmmoudM, @LauraT, @ElenaK, @HelenaB, @RobO, @ShafaqA, @Nancy A.
Day 21: October 14, 2019
I'm watching all the video lectures of the extracurricular material.
I'm in Part 2. Review: Training A Neural Network (Progress 16%); Lesson 1 - Feedforward and Backpropagation (Progress 100%); 5. Backpropagation
I applied to the Bertelsmann Technology Scholarship Program. I selected the Artificial Intelligence track. See the photo.
I encourage @KhushkumarP, @Jacqueline SusanM, @labibaR, @OudarjyaS, @MuhammadN, @Mohamed ChoukriB, @KonstantinosK, @AdityaK, @MahmmoudM, @LauraT, @ElenaK, @HelenaB, @RobO, @ShafaqA, @Nancy A.
Day 22: October 15, 2019
I'm watching all the video lectures of the extracurricular material.
I'm in Part 2. Review: Training A Neural Network (Progress 41%); Lesson 2 - Training Neural Networks (Progress 50%); 8. Dropout
I encourage @KhushkumarP, @Jacqueline SusanM, @labibaR, @OudarjyaS, @MuhammadN, @Mohamed ChoukriB, @KonstantinosK, @AdityaK, @MahmmoudM, @LauraT, @ElenaK, @HelenaB, @RobO, @ShafaqA, @Nancy A.
Day 23: October 16, 2019
I was selected as technical moderator of computer vision in the Slack channel #cvnd_techhelp of the Phase 2 of the Secure & Private Artificial Intelligence Challenge (SPAIC). See the photo.
I applied to become a mentor of the Computer Vision nanodegree. crossing fingers
I'm watching all the video lectures of the extracurricular material.
I'm in Part 2. Review: Training A Neural Network (Progress 75%); Lesson 3 - Deep Learning with PyTorch (Progress 27%); 3. PyTorch Tensors
I encourage @KhushkumarP, @Jacqueline SusanM, @labibaR, @OudarjyaS, @MuhammadN, @Mohamed ChoukriB, @KonstantinosK, @AdityaK, @MahmmoudM, @LauraT, @ElenaK, @HelenaB, @RobO, @ShafaqA, @Nancy A.
Day 24: October 17, 2019
I'm watching all the video lectures of the extracurricular material.
I'm in Part 2. Review: Training A Neural Network (Progress 82%); Lesson 3 - Deep Learning with PyTorch (Progress 45%); 5. Training Networks
I encourage @KhushkumarP, @Jacqueline SusanM, @labibaR, @OudarjyaS, @MuhammadN, @Mohamed ChoukriB, @KonstantinosK, @AdityaK, @MahmmoudM, @LauraT, @ElenaK, @HelenaB, @RobO, @ShafaqA, @Nancy A.
Day 25: October 18, 2019
I'm watching all the video lectures of the extracurricular material.
I'm in Part 2. Review: Training A Neural Network (Progress 88%); Lesson 3 - Deep Learning with PyTorch (Progress 64%); 7. Inference & Validation
I encourage @KhushkumarP, @Jacqueline SusanM, @labibaR, @OudarjyaS, @MuhammadN, @Mohamed ChoukriB, @KonstantinosK, @AdityaK, @MahmmoudM, @LauraT, @ElenaK, @HelenaB, @RobO, @ShafaqA, @Nancy A.
Day 26: October 19, 2019
I'm watching all the video lectures of the extracurricular material.
I'm in Part 2. Review: Training A Neural Network (Progress 94%); Lesson 3 - Deep Learning with PyTorch (Progress 82%); 9. Loading Data Sets with Torchvision
I'm watching all the video lectures of the extracurricular material.
I'm in Part 2. Review: Training A Neural Network (Progress 100%); Lesson 3 - Deep Learning with PyTorch (Progress 100%); 11. Transfer Learning Solution
I'm watching all the video lectures of the extracurricular material.
I'm still programming my vectorial field of rewards for AWS DeepRacer. The deadline for the final race is close.
I'm reading some links about fully convolutional neural networks and semantic segmentation, given that there is a lesson about it in the extracurricular section. See the photo.
I answered some questions in the Slack channel #cvnd_techhelp.
I'm watching all the video lectures of the extracurricular material.
Today I extensively investigated how to draw vectorial fields in matplotlib.
I programmed the first version of my vectorial field of rewards for AWS DeepRacer. However, my vectorial field of rewards is not aligned, yet, as you can see in the photo. The deadline for the final race of AWS DeepRacer is close. I need to hurry up!
I answered some questions in the Slack channel #cvnd_techhelp.
I'm watching all the video lectures of the extracurricular material.
The deadline for the final race of AWS DeepRacer is close. I need to hurry up!
I succesfully implemented my vectorial field of rewards for AWS DeepRacer. See the first photo.
I turned the steering wheel to make the car stay on track. Now my vectorial field of rewards looks like a centipede. I'm a mathematical artist. See the second photo.
I closed the loop of the track. That wasn't trivial. In fact, I still get errors in the AWS DeepRacer server. See the third photo.
I answered some questions in the Slack channel #cvnd_techhelp.
I'm watching all the video lectures of the extracurricular material.
The deadline for the final race of AWS DeepRacer is close. I need to hurry up!
Now I implemented my vectorial field of rewards for AWS DeepRacer by using only the 2 indexes provided by params['closest_waypoints']. My vectorial field is smooth and continuous. See the photo. I hope to test it soon in my first AWS DeepRacer submission.
I answered some questions in the Slack channel #cvnd_techhelp.
The deadline for the final race of AWS DeepRacer is close. I need to hurry up!
I'm training a new model whose maximum speed is 10 m/s. Training at high speeds takes a much longer time. After 12 hours of GPU training, my new model starts to show signs of understanding the problem of driving at high speeds. See the reward graph. That's why I'm creating a new kind of neural network to overcome the limitations of deep learning.
I answered some questions in the Slack channel #cvnd_techhelp.
The deadline for the final race of AWS DeepRacer is really close, in just 13 hours and a half. I need to hurry up!
I corrected 2 mistakes in my reward function for DeepRacer.
First, I forgot to take the absolute value of the steering angle to compute the steering factor. It was like driving with someone constantly pulling the steering wheel to one side. In spite of this inconvenience, Deep Reinforcement Learning managed to adapt to this deformed causal geometry and achieved great times.
Second, I forgot to convert to radians when computing the maximum steering angle. This error almost nullified the action of the steering factor because such factor was about 0.00436.
By programming a Python script, I automated the generation of messages to remind my Udacity students of their deadlines of the Intro to Self-Driving Cars nanodegree. See the photo.
After the stress caused by AWS DeepRacer, now I'm watching all the video lectures of the extracurricular material of the computer vision nanodegree.
I answered some questions in the Slack channel #cvnd_techhelp.
Day 40: November 2, 2019. #30DaysofUdacity #PracticeMakesPerfect
I'm watching all the video lectures of the extracurricular material of the computer vision nanodegree.
I'm in Extracurricular Part 3. Skin Cancer Detection (Progress 28%), Lesson 1 - Deep Learning for Cancer Detection with Sebastian Thrun (Progress 28%), 8. Solution: Data Challenges.
I answered some questions in the Slack channel #cvnd_techhelp.
Day 41: November 3, 2019. #30DaysofUdacity #PracticeMakesPerfect
I'm watching all the video lectures of the extracurricular material of the computer vision nanodegree.
I'm in Extracurricular Part 3. Skin Cancer Detection (Progress 100%), Lesson 1 - Deep Learning for Cancer Detection with Sebastian Thrun (Progress 100%), 29. Mini Project: Dermatologist AI.
I answered some questions in the Slack channel #cvnd_techhelp.
Day 45: November 7, 2019. #30DaysofUdacity #PracticeMakesPerfect
I'm watching all the video lectures of the extracurricular material of the computer vision nanodegree.
I talked to a friend in Finland, Karoliina Salminen, about the Izhikevich's model of spiking neurons. We are planning to implement it as an artificial intelligence project for the Phase 2 of SPAIC, in the Slack channel #sg_neuroscience. See the photo.
I answered some questions in the Slack channel #cvnd_techhelp.
Day 46: November 8, 2019. #30DaysofUdacity #PracticeMakesPerfect
I'm watching all the video lectures of the extracurricular material of the computer vision nanodegree.
Today I received my badge for completing the #30DaysofUdacity. See the photo.
It seems that I was selected to become a mentor of the computer vision nanodegree. I'm so happy. Thank Vienna Harvey
and Ravi Jain. I won't disappoint you. I'll do my best.
Today I took the computer-based GRE at Universidad Espíritu Santo. The GRE lasted 5 hours. I'm exhausted. I obtained 152/170 in Verbal Reasoning and 160/170 in Quantitative Reasoning, not bad. I took the GRE because I will apply for the Ph.D. program in Machine Learning at Georgia Institute of Technology, this December.
I answered some questions in the Slack channel #cvnd_techhelp.
Today, I watched the YouTube "Michael Bronstein - Geometric deep learning on graphs: going beyond Euclidean data", in the following link: https://www.youtube.com/watch?v=b187J4ndZWY