Recent developments in AI have made it ubiquitous, every industry is trying\nto adopt some form of intelligent processing of their data. Despite so many\nadvances in the field, AIs full capability is yet to be exploited by the\nindustry. Industries that involve some risk factors still remain cautious about\nthe usage of AI due to the lack of trust in such autonomous systems.\nPresent-day AI might be very good in a lot of things but it is very bad in\nreasoning and this behavior of AI can lead to catastrophic results. Autonomous\ncars crashing into a person or a drone getting stuck in a tree are a few\nexamples where AI decisions lead to catastrophic results. To develop insight\nand generate an explanation about the learning capability of AI, we will try to\nanalyze the working of loss functions. For our case, we will use two sets of\nloss functions, generalized loss functions like Binary cross-entropy or BCE and\nspecialized loss functions like Dice loss or focal loss. Through a series of\nexperiments, we will establish whether combining different loss functions is\nbetter than using a single loss function and if yes, then what is the reason\nbehind it. In order to establish the difference between generalized loss and\nspecialized losses, we will train several models using the above-mentioned\nlosses and then compare their robustness on adversarial examples. In\nparticular, we will look at how fast the accuracy of different models decreases\nwhen we change the pixels corresponding to the most salient gradients.\n