A few months ago I asked should quants learn more about machine learning?
Last month Google acquired DeepMind Technologies with the interest in Deep Learning. It seems that in the view of Google deep learning deals with the use of neural networks to build powerful general purpose learning algorithms.
Some writers distinguish deep learning from machine learning, especial because its was less explicit and not supervised, but unsupervised.
In short: in supervised learning data samples have an input and an output vector (goal parameters). Typical methods are Bayesian statistics, decision tree learning, neural networks, kernel methods, like support vector machines, inductive logic programming, (fuzzy) rule based learning, … Unsupervised learning extracts structure from unlabeled data. Typical methods are (fuzzy)clustering, self organizing maps, …. IMO, both are machine learning.
Machine learning in general is of bad nature for generalization. One of the approaches to generalize is, IMO, deep learning. Deep learning in neural networks is a special discipline.
Simplifying, deep learning seeks models "parameters" that are independent of partitions - by abstraction.
In quant finance we have (parametrized) abstractions: models and they are even calibrated to market data. But to build an adaptive technology would need much more information more frequently than our models for the movements of our underlings have.
Call it adaptive re-calibration or deep learning. The challenges are the same. We need sound mathematical schemes for valuations and informative data. The methodologies are known.
Picture from sehfelder
