Up to 70% of the total cost for AI development falls on labeling data and computing power. There are two main problems for AI - not enough labeled data to train neural networks, and expensive computational power which is a bottleneck. Neuromation aims to solve these problems by creating a marketplace platform that brings together AI practitioners and enterprise demand. It provides synthetic labeled datasets, distributed computing power, and machine learning models. Neuromation's utility token NTK will be the means of exchange on the platform.
2. BITCOIN OR
ETHER MINING
AMAZON DEEP
LEARNING
$7-8 USD
per DAY
$3-4 USD
HOUR
Chris
Up to 70% of the total cost for AI development
falls on labeling data and computing power
1. NOT ENOUGH LABELED DATA
2. EXPENSIVE COMPUTATIONAL POWER
THE TWO MAIN PROBLEMS OF AI
PROBLEM STATEMENT
3. BITCOIN OR
ETHER MINING
AMAZON DEEP
LEARNING
$7-8 USD
per DAY
$3-4 USD
HOUR
Chris
NOT ENOUGH LABELED DATA
TO TRAIN A NEURAL NETWORK
BOTTLENECK OF AUTOMATION
FOR EVERY INDUSTRY:
1 BLN LABELED PHOTOS = $240 MLN if processed manually
THE DATA PROBLEM
4. ● Labeled data with 100% accuracy
● Automated data generation with no limits
● Cheaper and faster than manual labeling
SYNTHETIC DATA:
BREAKTHROUGH IN DEEP LEARNING
BUT SYNTHETIC DATA REQUIRES EXPENSIVE COMPUTATIONAL POWER,
BOTH TO RENDER DATA AND TO TRAIN NEURAL NETWORKS
THE DATA PROBLEM
5. BITCOIN OR
ETHER MINING
AMAZON DEEP
LEARNING
$7-8 USD
per DAY
$3-4 USD
HOUR
Chris
ESTIMATED 40-50% OF ANY AI START-UP’S BUDGET
IS ATTRIBUTED TO COMPUTING POWER
COMPUTING POWER IS
EXPENSIVE AND SCARCE
BOTTLENECK
FOR AI PRACTITIONERS:
THE COMPUTATIONAL PROBLEM
6. BITCOIN OR
ETHER MINING
AMAZON DEEP
LEARNING
$7-8 USD
per DAY
$3-4 USD
HOUR
ETHER MINING AMAZON DEEP
LEARNING
$10-12
per DAY
$6-8
per HOUR
The AI industry is ready to
pay miners for their
computational resources -
more than they can ever get
from mining Ether
KNOWLEDGE MINING IS MORE PROFITABLE.
DEEP LEARNING NEEDS DISTRIBUTED COMPUTING POWER
WE CAN BRIDGE THIS GAP
GPU
x6
GPU
THE COMPUTATIONAL PROBLEM
7. BITCOIN OR
ETHER MINING
AMAZON DEEP
LEARNING
$7-8 USD
per DAY
$3-4 USD
HOUR
NEUROMATION
MARKETPLACE PLATFORM —
SOLVING BOTH PROBLEMS:
1. SYNTHETIC DATA
2. COMPUTING POWER
+
3. MACHINE / DEEP LEARNING MODELS
DISTRIBUTED
COMPUTING
POWER
SYNTHETIC
DATA SETS
MACHINE
LEARNING
MODELS
WE BRING TOGETHER AI PRACTITIONERS AND ENTERPRISE DEMAND
ON ONE INTEGRATED MARKETPLACE PLATFORM
NEUROMATION MARKETPLACE
8. Neurotoken (NTK)
NTK — the utility token for an
operating digital economy
● Neuromation Platform is a hub for the
global AI economy
● Tokens are the unit of exchange within
the Neuromation Platform ecosystem
● NTK liquidity is necessary for seamless
execution of transactions by all
participants
NTK IS THE MEANS OF EXCHANGE ON THE NEUROPLATFORM
WE HAVE SOLD OUT NTK IN 9 HOURS OF TOKEN SALE (+PRE-SALE)
TOKEN ECONOMICS
10. Chris
● Image recognition for retail:
automatically recognize goods on
the shelf with deep neural networks
● Central problem in many
applications of neural networks:
where do we get labeled data from?
LOTS OF LABELED DATA REQUIRED
IMAGE RECOGNITION ON THE SHELF
11. ● Current catalogue of goods only in
Russian-speaking (ex-USSR) retail:
● Each SKU requires
● Moreover, the network needs “real-looking”
photos, i.e., labeled photos ofreal shelves...
170 000 SKU
1000 - 5000
labeled photos
WHERE WILL ALL THIS LABELED DATA COME FROM?
IMAGE RECOGNITION ON THE SHELF
12. ● We create a virtual copy of the shelf (a 3D model) and place 3D objects
● Then we can generate unlimited amounts of labeled synthetic data
● We train deep neural networks on synthetic data
● Reaching good accuracy much faster and cheaper than the competition
IMAGE RECOGNITION ON THE SHELF
13. Advantages of
synthetic data
● We generate 100% accurate labeled
data, with pixel-perfect labeling
which is impossible to do by hand
● Increase the speed of automation by
orders of magnitude
● The resulting solution is several
times cheaper than hand labeling
IMAGE RECOGNITION ON THE SHELF
16. .
Piglet’s Big Brother
Piglet’s Big Brother
● Feeding is of paramount
importance in pig farming
● But farmers weigh pigs only twice
in their entire lives! Why?..
17. .
Piglet’s Big Brother
Piglet’s Big Brother
● Feeding is of paramount
importance in pig farming
● But farmers weigh pigs only twice
in their entire lives! Why?..
19. .
Piglet’s Big Brother
Piglet’s Big Brother
● Next step: video analysis to find
sick animals
● Not limited to pigs, of course
20. .
Piglet’s Big Brother
Our Team
Maxim Prasolov
CEO
Fedor Savchenko
CTO
Sergey Nikolenko
Chief Research Officer
Denis Popov
Chief Information
Officer
Constantine Goltsev
Investor / Chairman
Andrew Rabinovich
Adviser
Yuri Kundin
COO
David Orban
Adviser
Esther Katz
VP Communication
David Orban
Adviser