Hello! We apologize for the connection difficulty we had with Dr. Chai. We didn't get a chance to ask him these questions because of the connection issue, but here they are! Here is a brief explanation from him for some of the questions asked. If there are any more questions, you can email us at ftc18225@gmail.com or message us on Instagram at @ftc18225. Are there any disadvantages to deep learning and non-deep learning? - There may be overfeeding. Things are based on predictions and it can mess them up. Another disadvantage is that the prediction may be unexplainable--for example in medical instances when they find tumors or things like that, they need an explanation. Which layer of learning do you think is the most important? - Definitely the prediction layer. It goes very deep and is overall more complex. The convolution layer is more commonly used and generally more simple than the prediction layer.
Hello! We apologize for the connection difficulty we had with Dr. Chai. We didn't get a chance to ask him these questions because of the connection issue, but here they are! Here is a brief explanation from him for some of the questions asked. If there are any more questions, you can email us at ftc18225@gmail.com or message us on Instagram at @ftc18225.
Are there any disadvantages to deep learning and non-deep learning?
- There may be overfeeding. Things are based on predictions and it can mess them up. Another disadvantage is that the prediction may be unexplainable--for example in medical instances when they find tumors or things like that, they need an explanation.
Which layer of learning do you think is the most important?
- Definitely the prediction layer. It goes very deep and is overall more complex. The convolution layer is more commonly used and generally more simple than the prediction layer.
For our past Connecting with Professionals events please head over to ua-cam.com/channels/Kenlz_JmUq6KS9YrkgNT8g.html !