Whenever we talk about a 'single-node cluster' from a learning point of view, a 'pseudo-distributed mode' is a single-node cluster. This essentially means that the NameNode and DataNode (Master-Slave) are situated on the same machine.
This also means that there is no consideration for a wide resource management system spread throughout many users as there is only one user in the implementation. This sort of configuration is only used for learning or testing purposes.
In pseudo-distributed mode, every service runs on the same physical machine. Such configuration is mainly used while testing when we don’t need to think about the resources and other users sharing the resource. Every Hadoop component that is virtualized or used in this sort of architecture has a separate JVM created for it and they are interactive with the help of network sockets, creating a sort Hadoop cluster on a small scale without the use of multiple physical nodes.
It would really be helpful for you to get formal hadoop training or hadoop certification if you're getting into the field of big data if you wanna get started with the basic concepts.