How do you do deep learning research?
Search for research inspiration in different places
- Talk to a researcher in a different field. Ask what problem they are excited about and try to restate the problem in computational terms.
- Code a simple baseline to get a feel for a problem.
- Extend the experiments section of a paper you like.
What specs are needed for machine learning?
Configuration of the Workstation:
- Processor – Intel Xeon E2630 v4 – 10 core processor, 2.2 GHz with Turboboost upto 3.1 GHz.
- Motherboard – ASRock EPC612D8A.
- RAM – 128 GB DDR4 2133 MHz.
- 2 TB Hard Disk (7200 RPM) + 512 GB SSD.
- GPU – NVidia TitanX Pascal (12 GB VRAM)
- Intel Heatsink to keep temperature under control.
What do machine learning researchers do?
Machine learning researchers or data scientists are people who work with data and build machine learning models. They clean and interpret data and build models using a combination of machine learning algorithms and data. Simply put, university researchers often work alone or in small teams.
What are the research areas in AI?
8 Best Topics for Research and Thesis in Artificial Intelligence
- Machine Learning.
- Deep Learning.
- Reinforcement Learning.
- Robotics.
- Natural Language Processing.
- Computer Vision.
- Recommender Systems.
- Internet of Things.
What is the learning task in reinforcement learning?
Reinforcement learning is a machine learning training method based on rewarding desired behaviors and/or punishing undesired ones. In general, a reinforcement learning agent is able to perceive and interpret its environment, take actions and learn through trial and error.
Is neural network a reinforcement learning?
Neural networks are function approximators, which are particularly useful in reinforcement learning when the state space or action space are too large to be completely known. A neural network can be used to approximate a value function, or a policy function.
What are the best resources to learn reinforcement learning?
Once you have got a good hang of basic reinforcement learning concepts, start following lectures from UC Berkeley Deep Reinforcement Learning course and David Silver’s lectures on Reinforcement Learning.
Is it advisable to jump right into deep reinforcement learning?
Jumping right into Deep Reinforcement Learning is not advisable if you only understand Deep Learning part and not the Reinforcement Learning part. That’s one major fallacy of folks who are pretty well versed in Deep Learning but have no idea what Reinforcement Learning is about.
How important is the Epsilon in reinforcement learning?
As you start to play around with Reinforcement Learning problems, you will start to realize how brittle the parameters are. Tuning your epsilon to a particular number to have enough exploration done before your agent starts exploiting is as important as setting up an exact architecture with exact parameters for your DQN network.