Ladder Networks: Learning under Massive Label Deficit
Journal Title: International Journal of Advanced Computer Science & Applications - Year 2017, Vol 8, Issue 7
Abstract
Advancement in deep unsupervised learning are finally bringing machine learning close to natural learning, which happens with as few as one labeled instance. Ladder Networks are the newest deep learning architecture that proposes semi-supervised learning at scale. This work discusses how the ladder network model successfully combines supervised and unsupervised learning taking it beyond the pre-training realm. The model learns from the structure, rather than the labels alone transforming it from a label learner to a structural observer. We extend the previously-reported results by lowering the number of labels, and report an error of 1.27 on 40 labels only, on the MNIST dataset that in a fully supervised setting, uses 60000 labeled training instances.
Authors and Affiliations
Behroz Mirza, Tahir Syed, Jamshed Memon, Yameen Malik
Keywords
Related Articles
- EP ID EP260686
- DOI 10.14569/IJACSA.2017.080769
- Views 90
- Downloads 0
A Brief Survey on 5G Wireless Mobile Network
The new upcoming technology of the fifth generation wireless mobile network is advertised as lightning speed internet, everywhere, for everything, for everyone in the nearest future. There are a lot of efforts and resear...
Improving Vertical Handoffs Using Mobility Prediction
The recent advances in wireless communications require integration of multiple network technologies in order to satisfy the increasing demand of mobile users. Mobility in such a heterogeneous environment entails that use...
Seamless Connectivity for Adaptive Multimedia Provisioning over P2P-enabled IP Multimedia Subsystem
The subsystem multimedia internet network (IMS) has been upgraded to support peer-to-peer content distribution services, the peer heterogeneity constraint imposes a challenge that is the guarantee of a certain level of Q...
Pipeline Hazards Resolution for a New Programmable Instruction Set RISC Processor
The work presented in this paper is a part of a project that aims to concept and implement a hardwired programmable processor. A 32-bit RISC processor with customizable ALU (Arithmetic and Logic Unit) is designed then th...
Reduced Complexity Divide and Conquer Algorithm for Large Scale TSPs
The Traveling Salesman Problem (TSP) is the problem of finding the shortest path passing through all given cities while only passing by each city once and finishing at the same starting city. This problem has NP-hard com...