While Machine Learning is a wonderful tool for high abstraction-level problems, we do not think it should be considered as the only or definitive solution for most of the problems related to Situation Awareness, because of several serious drawbacks:
- We don’t really know why Neuronal Networks make the choices they do. This black-box nature makes it much harder (yet not impossible) for these systems to meet functional safety requirements like those in the ISO26262 standard.
- Machine Learning is far from being infallible: Besides the problems of overfitting and under-fitting, Neural Networks operate in an opaque way. Several studies have revealed that artificial perturbations on natural images can easily make DNN (Deep Neuronal Networks) misclassify objects and hamper its functionality through the insertion of effective algorithms to generate alternate samples called “adversarial images”.
Two of the most known examples of this are the one-pixel attack and the hacking that tricked a Tesla into veering into the wrong lane.
- The Machine Learning process has to go through too much useless data: Even if the inference phase is much lighter in terms of processing requirements than the training phase, it still requires the input of the full raw data from sensors.
This huge amount of information takes up valuable network bandwidth, processing power and/or costs too much with regards to time and energy consumption.
- The Machine Learning approach relies heavily on the invisible labor of humans, who must tediously label the training data for it to work. These people are often working in isolated and hard conditions.
As sensors evolve, the data labelling needs to be updated. This titanic effort is also an endless one.
The mainstream approach of the industry is to make extensive use of Machine Learning methods, relying on finding patterns in trained datasets to infer qualitative (statistical) conclusions on new samples.
Our heretical approach is not to use Machine Learning. No training, no datasets, no labelling. We instead rely on deductive inference to get to quantitative conclusions.
We feed the central decision-making processes with edge-computed and already classified data (ie. only meaningful and relevant information).
Machine Learning is not the silver bullet that some purport it to be. It is not the only, or even the best technology to solve every problem, yet it can be incredibly useful on some high-abstraction level ones.
Most Situation Awareness' problems are not nails, so choose your tool wisely!