their is always some uncertainity involved in AI.
this uncertainity comes from:
- insufficient data
- noise in data
- errors while collecting data.
- assumptions when modelling
we can represent this uncertainity qualitatively with the mathematical theory of probalility and statistics.
this provides the foundation and tools for quantifying, handling and harnessing uncertainity
this provides us with the methods of collecting presenting analysing intepreting and inferencing from data.
With all these tools, we are equipped to mathematically define decision- making, which is required to automate decision-making from data, that is, to achieve the final goal of AI.
this decisions can be of two types:
- discrete
- continuous
Mathematically, discrete decisions can be represented as a way of partitioning the high dimensional space where the data points lie and assigning a category to each partition. Continuous decisions, on the other hand, are some functions mapping a point in high dimensional space to a real number.
- Linear Algebra
- Vector Calculus
- Probability
- Statistics