BOSTON, April 29, 2016 (GLOBE NEWSWIRE) -- indico, an innovator in the machine learning and artificial intelligence space, will be presenting a research paper in the poster presentation track at the 2016 International Conference on Learning Representation (ICLR) taking place next week in San Juan, Puerto Rico, May 2-4.
Alec Radford, co-founder and head of research, and Luke Metz research engineer at indico, will present the paper - Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks - along with Soumith Chintala, artificial intelligence research engineer at Facebook.
ICLR, is an annual conference sponsored by the Computational and Biological Learning Society. The 2016 event, chaired by Yoshua Bengio and Yann Lecun, focuses on the important connections between representation learning and machine learning and to application areas such as vision, speech, audio and NLP. The conference takes a broad view of the field, and includes topics such as deep learning and feature learning, metric learning, kernel learning, compositional models, non-linear structured prediction, and issues regarding non-convex optimization.
ICLR Poster Presentation Overview
|Alec Radford||Co-founder and head of research, indico|
|Luke Metz||Research engineer, indico|
|Soumith Chintala||Artificial intelligence research engineer, Facebook|
In recent years, supervised learning with convolutional networks (CNNs) has seen huge adoption in computer vision applications. Comparatively, unsupervised learning with CNNs has received less attention. In this work we hope to help bridge the gap between the success of CNNs for supervised learning and unsupervised learning. We introduce a class of CNNs called deep convolutional generative adversarial networks (DCGANs), that have certain architectural constraints, and demonstrate that they are a strong candidate for unsupervised learning. Training on various image datasets, we show convincing evidence that our deep convolutional adversarial pair learns a hierarchy of representations from object parts to scenes in both the generator and discriminator. Additionally, we use the learned features for novel tasks - demonstrating their applicability as general image representations.
indico provides state-of-the-art machine learning algorithms for text and image analysis in the form of a simple to use web service. This, for the first time, enables companies to automatically extract meaningful insight from unstructured data regardless of their size or capability. Sentiment Analysis, Social Media Monitoring, Content Filtering, Content Classification, Recommendation, and Personalization are just some of the areas in which indico’s customers are deploying its technology to improve business outcomes. Furthermore, indico’s rapid customization capabilities have also enabled companies such as Mavrck, CO Everywhere, and interlinkONE to quickly develop compelling new solutions that weren’t practical before. indico is privately held and headquartered in Boston, MA. For more information, visit https://indico.io/.