. . When can we achieve 100% training accuracy for Multi-layer perceptron ? Solution: In general, it is not always possible to achieve 100% training accuracy for a Multi-Layer Perceptron (MLP) on all datasets, particularly when dealing with complex, real-world problems. However, for simple and linearly separable datasets, it may be possible to achieve 100% training accuracy with an MLP. This is because the decision boundary between classes is a straight line or a hyperplane, which can be learned by a single-layer MLP. This is because the decision boundary between classes is a straight line or a hyperplane, which can be learned by a single-layer MLP