|M.Sc Student||Shoshan Alon|
|Subject||Dynamic-Net: Tuning the Objective without Re-Training|
for Synthesis Tasks
|Department||Department of Electrical and Computer Engineering||Supervisors||PROF. Ayellet Tal|
|PROF. Lihi Zelnik-Manor|
One of the key ingredients for successful optimization of modern CNNs is identifying
a suitable objective. To date, the objective is fixed a-priori at training time, and any
variation to it requires re-training a new network. In this paper we present a first attempt at alleviating the need for re-training. Rather than fixing the network at training time, we train a “Dynamic-Net” that can be modified at inference time. Our approach considers an “objective-space” as the space of all linear combinations of two objectives, and the Dynamic-Net is emulating the traversing of this objective-space at test-time, without any further training. We show that this upgrades pre-trained networks by providing an out-of-learning extension, while maintaining the performance quality. The solution we propose is fast and allows a user to interactively modify the network, in real-time, in order to obtain the result he/she desires. We show the benefits of such an approach via several different applications.