Deep Learning, 5 blunt pros and cons.

PROS:

1. In deep learning, the features can be automatically extracted and tuned for the necessary outcome.

2. These features need not be extracted well ahead of execution time. This can prevent wastage of time like that in the traditional ML algorithms.

3. The DL or DNN architectures are flexible and adaptable to newer examples and problems.

4. Neural Network based approcah can be applied to many different applications and data types.

5. DL can perform the parallel computation with the help of GPUs.

CONS:

1. Deep Learning requires extremely large amount of data in order to achieve a fairly good accuracy.

2. It is not easy to comprehend output based on only the learning and it requires classifiers to do so.

3. Extremely expensive to train due to complex data models.

4. DL models are difficult to be adopted by less skilled people as there is no standard guide for the selection of the exact tools and suitable architecture. Also, a prior knowledge is required w.r.t. the hyper parameters and the knowledge of the architectures.

5. A large number of parameters are required to be trained and that is why DL is considered time consuming and computationally intensive.