June 18: Guest Speaker

 In today's class we had a guest speaker. The guest speaker was Dr. Schuman. She discussed machine learning and talked about some of her projects at UT. 

Machine Learning: What Is Machine Learning? - TechUpdatesDaily

Summary:

Dr. Schuman discussed many things in her presentation. One of the main things was how a machine learns. She discussed different methods of this learning. The one that stuck with me the most was the K- nearest neighbor method. In simple terms the method uses whatever "neighbor" is nearest to the point you selected and used that to assume what the object should be. The machine declares the object whatever neighbor is most prominent near it. The number of "neighbors" used to determine the new object can be changed depending on how accurate things are spaced or in general need to be. People typically use odd number for K-nearest neighbor to avoid ties. Besides the machine learning, she also showed us one of her projects at UT. It was a small car that was being taught to drive itself. The car used machine learning to navigate paths. She said the car can get up to 40mph, but to drive without hitting obstacles she says her goal is to get it up to 15mph. She showed a video of the car moving through a makeshift course and we got to see the car learning and navigating itself. The car moved in a side-to-side movement to analyze its surroundings. The reason Schuman gave us for why the car cannot travel fast through the course is the cord they are using. Because there using a USB cord the time it takes for the car to receive and respond to its surroundings is slowed. 

What I learned: 

I found today's presentation to be really interesting. I definitely think I can apply this somewhere outside of GSSE. I am a part of a First robotics team. In First each competition starts off with a 30 sec autonomous period. I think this information about machine learning could help us improve how we code our robot depending on the game for next year. I would love to learn more and see how I could apply machine learning to next year's robotics season. Since these competitions are done in teams our position is never the same on the field. If the robot could learn from where it is to shoot on its own in the 30 period it would benefit our teamwork with our alliance as we would be able to start from anywhere rather than having to have a fixed position to start on. 



Comments

Post a Comment

Popular posts from this blog

June 10 3d-printing

Image
 Today in class we learned about 3d printing and made our own Cad designs to print off.    Onshape | Product Development Platform I myself have not done any 3d Printing however my robotics teams design team uses it a lot. I know that for a 3d print you must first create your model. Creating a model on a CAD program I have done before. I have used Tinkercad a little bit and more recently I learned Onshape for my engineering class. I choose Onshape as I have the most experience with it and I personally like it better than Tinkercad.   My design: I designed a tablet for my DND dice for when I am playing. I designed it to keep my dice in place as I tend to lose my dice very easily. I made it in the shape of a sword to reflect on the fantasy of DND and also just because I thought it looked cool.  What I learned: I learned more about the printing side of 3D models. While I am used to making CAD models and seeing the outcome of 3D printing I have never actually seen how it is done and the rul
Read more

Visualization and AI Overview.

Image
Today's blog will be about two different things. First Is about data visualization and the second is about AI which we used to guess who would survive on the titanic.  Data visualization: I was not able to get excel to work on my computer, so I moved on to the python visualizations. I started with the plant visualizations. The first thing I figured out on it was how to move the planets and change the color. Later on, I learned how to change the shape of the planets and move the names around.  I learned from this how you can manipulate shapes and colors to portray what you want to more clearly. Color and Shape are just two of the ways to do data visualization.  Now let's take a look at AI: AI, also known as Artificial intelligence, is a program that learns from the data it is given to do a certain task. In class we discussed who would survive in the titanic. My group determined our answers from what class the people were and their age.  We assumed that people who were lower clas
Read more

June 4: Engineering Design Process

Image
Today we looked at the engineering design process and then applied it in a quick project. T he  project  was simply to  move  an  object  from one place to another. My group started our project like this:   Step 1(Define): We choose to move an eraser from one side of the table to the other.  Step 2(Ask): My group identified what forces we could work with to achieve our goal . We also identified the constraints of time and material .  Step 3(Imagine): We listed out multiple ideas ranging from a Slingshot to a car .  Step 4(Plan): Our group choose to make a rubber band powered car . We then drew out our design to see how we could build it.  Step 5(Create) : At this stage we began to build our prototype from the wheels up splitting the work between the three of us. Step 6(Test): Finally, once we finished our prototype, we tested our idea .   Engineering Design Process - TeachEngineering                                                        Final outcome:                  Our
Read more