3 ms·
This is identical to the problem blockchain solves - hard to solve, but easy to verify. For example, if the workload is training a neural network on a set of sa
by TrainedMonkey 9y ago
This is identical to the problem blockchain solves - hard to solve, but easy to verify. For example, if the workload is training a neural network on a set of samples it is relatively cheap verify performance on a training sample.
Alternatively you could give X% of tasks to multiple workers for cross checking. In your example X% is 100%, but it does not have to be that high.
- mattkrause 9y agoSay you do a single feedforward pass. How do you verify that its output is the result of a trained network-and particularly a completely trained one?
- TrainedMonkey 9y agoAside from doing the same work X% of the time I can think of only one other option. The network could be overloaded with additional nodes such that it would also compute some function on the input data with the known output. For example a hashing function. Having said that, single pass does not strike me as a particularly useful unit of work. You would want to batch the samples, and that leads back to a strategy of verifying a portion of them based on the host trust metric.