Deep learning is expensive and time-consuming. Image recognition, natural language processing, computer vision, etc..., all require extracting millions of parameters to identify structure and patterns. This is very computationally intensive and in conventional methods there is an inherent latency in the architecture.
We solve latency by chunking the data into really small pieces (bytes), maintaining its identity, and then distributing it across the host of devices with a call to action: gradient calculations.
Raven removes added dependency on the Model replication and keeps the Model is intact at the Master Node. The training of any deeper neural network and their large datasets are distributed over the network of contributors.
Data is also sharded in smaller snippets.
This makes it easier for calculations to pass through from machine to machine, rather than creating multiple replicas of a complicated Model.
Raven crowd-sources compute power using Data and Model Parallelisation distribution approaches. Security and anonymity is guaranteed while distributing the training across multiple devices over the Internet, by being set on a blockchain.
This also allows for new revenue opportunities for the contributors and partners; who are contributing to the ecosystem growth, in the form of a constant source of income from such DL training.
All the frameworks operate on tensors and are built on the computational graph as a Directed Acyclic Graph. In most of the current and popular deep learning frameworks the computational graph is static in nature.
Attention. There is a risk that unverified members are not actually members of the team
Cloud computing market is projected to reach $410B by 2020, according to Forbes and Gartner. More data stipulate additional demand for such services resulting in projected CAGR of about 18%. So, the market Raven Protocol enters is attractive and future-oriented. But it is monopolized now by giants like Microsoft, Amazon and Alibaba. These vendors are widely known and respected with high guarantees, but the prices are also high as well. The blockchain is a beautiful solution and decentralization here has some rational grounds. If blockchain startups offer high-level security and stability the market may transform significantly. Most of the projects are still in the development stage but are rated well by community resulting in high investment returns: SONM, Dadi, Golem.However, recently there were many decentralized cloud-computing ICOs launched so the market competition for Raven Protocol might be tough.Technical details of Raven are unavailable and the competitive advantage is not clearly stated.
Raven Protocol aims to create a network of computing nodes that utilize idle compute power for the purposes of AI training. Raven Protocol will use Dynamic Distribution to avoid any lapse in the model training. It will have No-Dependency Net to freely welcome Contributors. Unlike other projects where contributors have to install a software to start contributing. The protocol will be secured and tokenized to secure the data and transaction and incentivize contributors and believers.Now, the project is on a proof of concept stage. The public release of the code is scheduled for October'18. Alpha version will be launched in April'19. Testnet expected in July'19, Mainnet Launch expected in September'19.
According to rumours, Raven aims to sell 40% of the tokens for 20K-25K ETH(around $10 million if 1 ETH = 400 USD). Then, the company valuation will be equal to 55K-60K ETH ($22-24 million). The valuation looks reasonable and does not seem too high. 35% of the tokens will be reserved for Ecosystem Growth, 25% will be allocated to the company funds.However, the team does not explain how the money raised will be spent.The team will release an additional 10% of the tokens each month following the Token Generation Event. Thus 90% of the tokens will be released in 5 months after the TGE. This scheme may devaluate the token price during these 5 months.
There are 5 people on the core team. The CEO of Raven Sherman Lee has a background in Computer Science(BSc in CS from Berkley) and previously worked as a Software Engineer at Yahoo!, Ask.com and a Product Manager at Yahoo!. Later, he became a partner at Hong-Kong based VC fund Zeroth.ai and is still employed at the fund, according to LinkedIn. Other co-founders also have startup backgrounds and previously created Mate Labs (AI startup) and Rocco (AI-powered social media marketing tool).2 out of 5 claimed team members do not mention Raven as a workplace on LinkedIn. The team is scattered between Hong Kong and India.
New project on the radar. Raven Protocol is kind of projects that you rarely meet nowadays.There are no fancy advisors and funds involved. It is a community driven project with an engaged team of true decentralization enthusiasts.Raven is building a decentralized Deep Learning training protocol. Today, the process of training AI/ML models could take weeks and even months to complete while the costs for computing resources usually are far beyond that most of small companies and groups of developers can afford. To change this, Raven lowers entry barrier by reducing the compute power required from the contributor, thus the cost of acquiring specific powerful CPUs and GPUs to train DNNs becomes minimal. The concept of sharing idle computing power to facilitate training saves the enormous expense involved. In return, the contributors are compensated/rewarded with Raven tokens.The team is focusing on fast-paced AI training. For example, a 1M image dataset that takes 2-3 weeks to train on AWS will be reduced down to 2-3 hours on Raven. We want to highlight the team behind the raven, these guys are true enthusiasts who are running the tech forward. Raven is run by a small team of 8 people led by Sherman Lee a prominent thought leader in the blockchain space.While co-founder Kailash Ahirwar recently wrote a book on Generative Adversarial Networks ProjectsWe strongly advise keeping an eye on Raven Protocol updates in the upcoming weeks:)
First Binance Initial Dex offering. Supply of the only 3M$ could be Matic 2.0. Long term the token can have a real use case. I am certainly aiming for IEO.Short-term: Very good Long-term: Good ICO rating: 8.8/10
Decentralised & distributed deep-learning training protocol. First IDO on Binance DEX. №1 challenger to be listed on Binance during the BEP-2 community listing. At current price of 0.0029$ hardcap is just 5 mln $.
This offer is based on information provided solely by the offeror and other publicly available information. The token sale or exchange event is entirely unrelated to ICOholder and ICOholder has no involvement in it (including any technical support or promotion). Token sales listed from persons that ICOholder has no relationship with are shown only to help customers keep track of the activity taking place within the overall token sector. This information is not intended to amount to advice on which you should rely. You must obtain professional or specialist advice or carry out your own due diligence before taking, or refraining from, any action on the basis of the content on our site. Any terms and conditions entered into by contributors in respect of the acquisition of Tokens are between them and the issuer of the Token and ICOholder is not the seller of such Tokens. ICOholder has no legal responsibility for any representations made by third parties in respect of any Token sale and any claim for breach of contract must also be made directly against the Token issuing entity listed herein.
If you have any concerns about the nature, propriety or legality of this token sale or the persons involved in it please contact [email protected] with detailed information about your concerns.