Table 2.

Summary of the comparative on the advantages and limitations of ANNs, RNNs, DRC, and OERC

Category	ANN	RNN	DRC	OERC
Processing Type	Static, feed-forward computation	Sequential processing with feedback loops	Temporal via fixed reservoir	Parallel, optoelectronic feedback
Nonlinearity Source	Activation functions	Activation + recurrent feedback	Reservoir node interactions	Optical nonlinearities
Memory / Temporal Dynamics	None	Temporal memory	Short-term memory	Tunable temporal response
Training Complexity	Full backpropagation	Backpropagation through time	Only readout layer trained (simple linear regression)	Only readout layer trained
Training Efficiency	High computational cost for deep networks	High due to sequential dependency	Moderate; depends on reservoir size	High; optical domain enables low latency and energy efficiency
Scalability	High in digital systems	Limited by training time	Moderate scalability in digital hardware	High with photonic integration
Main Advantages	Simple and versatile for static data	Strong sequential data handling	Easy training, temporal processing	High bandwidth, low power, real-time analog computation
Main Limitations	No time processing	Training instability and gradient issues	Limited precision	Fabrication complexity