Neuromation Q3 2019 Update

Neuromation
Nov 25, 2019 · 5 min read

Following last quarter’s launch of the Neuromation Marketplace, Neuromation has expanded its offering of AI services, compute resources and synthetic data. Our AI specialists are standing by and looking forward to hearing from you. Neurotoken (NTK) can be used in the marketplace for AI services including data labeling, model development, synthetic data creation, AI strategy consulting and fundamental research. Marketplace users also continue to have the ability to rapidly and seamlessly provision cloud resources, services and storage for the training of ML models via our developer platform.

As described in the brief updates below, Neuromation has made significant progress in terms of improved functionality of the Neuromation Platform and in fundamental AI research.

During the quarter, we are also pleased to announce that we have added a key new member of our team. Mariya Davidova, has joined Neuromation as a Developer Advocate. Formerly with software development and project management tool creator JetBrains, Mariya’s new role encompasses product community management, hands on customer engineering support, technical writing and, most importantly, seeking out feedback from our developer community and working to make sure our platform is the best possible product for their needs.

Development Update

During the quarter, Neuromation’s development team made significant progress updating and improving the functionality of our core product, the Neuromation Platform. We are pleased to announce that the platform is now multi-cloud with the addition of AWS and GCP support, and our team is making progress towards our future goal of supporting all major cloud services plus on-prem, thus freeing developers and our customer base from vendor lock-in. Google TPU support was also added during the quarter and we are happy to announce that our Python API for the platform is now public.

Our development team also made significant usability improvements to the platform during the quarter, including greatly improved storage operations and the ability to manipulate training environments by successfully implementing ‘saved job states’ on the platform. We also made incremental enhancements to our CLI and web interface, both according to our roadmap and in response to user feedback.

In order to ease the onboarding of new users to the platform and accelerate their time to productivity, the development team successfully implemented “Project Template” functionality during the quarter, which provides users with a recommended project structure makefile containing basic operations, accompanied by a predefined base environment to run user jobs. The beauty of this system is that it helps to solve the problem of manipulating docker images by successfully hiding all docker details under the hood, dramatically improving developer productivity.

Finally, we have also begun to roll out ML Recipes on the platform, which are crafted to enrich our users’ experience with ready-to-go solutions of various applied ML problems. We currently have 4 recipes available: object detection, collection of NLP models, a Fast.AI course and Fast.ai NLP course wrapped into our project templates. Going forward, we look forward to hearing feedback from users who are implementing these recipes on our platform and would also be very happy to help users create their own recipes and support those recipes on the platform going forward.

Research Update

During the quarter, the research team continued to do important work on fundamental and applied research; we are looking forward to a highly productive Q4 of 2019 and Q1 of 2020.

The results of our joint project with SolidOpinion have been published as an arXiv preprint, and the paper is already under review at a major conference. The Neuromation research team also initiated a new joint research project with the European Molecular Biology Lab (EMBL Heidelberg) working on quantifying colocalization between mass-spectrometry images based on deep learning models.

Our paper on breast tumor cellularity assessment was accepted to the ICCV 2019 workshop «Visual Recognition for Medical Images»; ICCV is one of the top AI conferences in the world, and we are proud to be presented there.

We also wrote and published a number of new NeuroNuggets blog posts, which are listed below.

New publications with Neuromation affiliation

  1. M. Panin, S.I. Nikolenko. Rendering Atmospheric Clouds with Latent Space Light Probes. Accepted to 12th ACM SIGGRAPH Conference and Exhibition on Computer Graphics & Interactive Techniques in Asia (SIGGRAPH ASIA 2019), 2019.
  2. S.I. Nikolenko, E. Tutubalina, Z. Miftakhutdinov, E. Beloded. CommentsRadar: Dive into Unique Data on All Comments on the Web. ArXiv preprint 1908.07069 [cs.IR], 2019
  3. K. Ovchinnikova, A. Rakhlin, L. Stuart, S.I. Nikolenko, T. Alexandrov. ColocAI: artificial intelligence approach to quantify co-localization between mass spectrometry images. bioRxiv preprint 10.1101/758425, 2019
  4. D. Nikulin, A. Ianina, V. Aliev, S.I. Nikolenko. Free-Lunch Saliency via Attention in Atari Agents. Accepted to ICCV 2019 Workshop «Interpreting and Explaining Visual Artificial Intelligence Models» (ICCVW 2019), 2019, arXiv:1908.02511 [cs.LG].
  5. I. Bochkov, A. Davydow, N. Gaevoy, S.I. Nikolenko. New Competitiveness Bounds for the Shared Memory Switch. ArXiv preprint 1907.04399 [cs.NI], 2019

New blog posts (NeuroNuggets)

  1. https://medium.com/neuromation-blog/neuronuggets-acl-in-review-i-3726e7a8c4e2
  2. https://medium.com/neuromation-blog/neuronuggets-acl-in-review-ii-7d6119697cbd
  3. https://medium.com/neuromation-blog/neuronuggets-acl-in-review-iii-fcc3b5418ad6
  4. https://medium.com/neuromation-blog/neuronuggets-acl-in-review-iv-596c6bcce65f

NTK Update

Q3 saw a significant price decline in the major crypto markets, led by ETH which finished the quarter down 41% and BTC down 25% in dollar terms. During the quarter, the total market cap of all major cryptocurrencies contracted by 38%. During this same period, NTK was down 24% in ETH terms, 41% in BTC terms and 56% in USD terms — and finished the quarter trading at $0.011348.

If we look at the cohort of the 2018 ICOs to which Neuromation belongs and which we have been tracking over previous quarters, the overall performance of the group was down 35%, but 78% of these tokens saw declining prices during the quarter and only four tokens on our list appreciated. If we look at only those tokens with negative price movement, the average decline in the quarter was 55%.

As we have stated numerous times in the past, the sole usage of NTK is for the purpose of purchasing AI development services. This is possible via the Neuromation Marketplace, which allows for NTK to be exchanged for various AI development services. We encourage all holders of NTK who are interested in such services to visit the Neuromation Marketplace website and to get in touch with our AI specialists to discuss your needs.

Neuromation

Distributed Synthetic Data Platform for Deep Learning Applications

Neuromation

Written by

https://neuromation.io

Neuromation

Distributed Synthetic Data Platform for Deep Learning Applications

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade