The contributions of this paper are as follows (Figure 1): 1.

Although it is often argued that this representational format is useful in learning to solve many real-world down-stream tasks, there islittle empirical evidence that supports this claim.

A central problem to understanding intelligence is the concept of generalisation. This allows previously learnt structure to be exploited to solve tasks in novel situations differing in their particularities.

The transformer architecture (V aswani et al., 2017) has revolutionized the field of machine learning,

Neural Information Processing Systems http://nips.cc/

These kinds of attacks are known as poisoning attacks .