3 ms·
Neural nets are a useful technology but they are not magic. I doubt they will play much of a role in any effort to improve financial markets. First, neural net
by spikels 13y ago
Neural nets are a useful technology but they are not magic. I doubt they will play much of a role in any effort to improve financial markets.
First, neural nets have been used for decades by various financial market participants, in particular hedge funds, and there impact has been very modest. Second, they are hard to understand thus unlikely to be favored most governments, regulators or even market participants in the unlikely event there was some agreement for global market reform. Lastly, there is no evidence that neural nets have any real advantage in valuing assets over existing techniques. If there were they would be more widely used for that purpose.
- ozymondias 13y agothe Neural Network as a fundamental device is only as powerful as the person who designed it. However you could add every single possible training data segment in as an input parameter, however if you don't have the horsepower and fast RAM storage to process the information then it will take an extremely long amount of time. with the advent of GPGPU technologies, neural nets are resurfacing as the goto machine learning algorithm.
- spikels 13y agoYes they are great and all the things you mention are super cool but since neural nets are only as good as the humans that design them we can't expect them to solve fundamental problems like market failures. For example would better weapons prevent wars? Maybe but not always and not forever?
- ozymondias 13y agoYou make good points, however it wouldn't be one or two networks that would solve such an incredibly complex problem. In my next post I'll try and answer this question, I still don't see eye to eye with you on this as I feel that a fairly well abstracted net structure could solve this.