MIT boffins develop a robot that could probably beat you at Jenga

A COLLECTIVE OF BOFFINS from the Massachusetts Institute of Technology (MIT) have developed a robot that can teach itself the complex physics of Jenga.

According to the researchers, the manipulator arm of the robot relies on a machine learning algorithm, visual data, and tactile feedback to teach itself how to correctly move blocks in the classic block-stacking game.

Playing Jenga requires a soft touch to prevent the tower from tumbling down. It also demands mastery of other skills, such as pushing, pulling, probing, placing, and aligning the blocks in the tower.

In its latest research, the team equipped an industrial ABB IRB 20 robotic arm with a soft-pronged gripper, an external camera, and a force-sensing wrist cuff. To train the machine, it was directed to randomly select a block in the Jenga tower and also select a location on that block to push and move it.

Each time the arm pushed the block, a connected computer recorded the force and visual measurements and compared those to previous moves before categorising the attempt as successful or unsuccessful.

Rather than performing thousands of such attempts, the robot was trained on just 300 moves. The team grouped attempts of similar measurements and outcomes in different clusters, each signifying specific block behaviour. Within around 300 attempts, the robot learned to intelligently anticipate its moves, guessing which blocks would be more difficult to move than others, and which might cause the tower to collapse.

The performance of the robotic arm was also compared with the performance of human players. The team found the success rate of the robot in keeping the tower upright while removing the wooden blocks was almost on a par with that of human players.

The team believes this technology could be useful in production units where delicate touch and careful vision are needed, for example, assembly of cellular phones or other smaller parts in a, separating recycling items from waste, etc.

“In a cellphone assembly line, in almost every single step, the feeling of a snap-fit, or a threaded screw, is coming from force and touch rather than vision,” says Alberto Rodriguez, Assistant Professor in the Department of Mechanical Engineering at MIT, and the lead researcher of the study.

“Learning models for those actions is prime real-estate for this kind of technology.”

The findings of the research are published in journal Science Robotics.

 


Source
Author: Inquirer Staff
Image Credit

A Faster, More Efficient Cryptocurrency

Design reduces by 99 percent the data users need to join the network and verify transactions.

 

MIT researchers have developed a new cryptocurrency that drastically reduces the data users need to join the network and verify transactions — by up to 99 percent compared to today’s popular cryptocurrencies. This means a much more scalable network.

Cryptocurrencies, such as the popular Bitcoin, are networks built on the blockchain, a financial ledger formatted in a sequence of individual blocks, each containing transaction data. These networks are decentralized, meaning there are no banks or organizations to manage funds and balances, so users join forces to store and verify the transactions.

But decentralization leads to a scalability problem. To join a cryptocurrency, new users must download and store all transaction data from hundreds of thousands of individual blocks. They must also store these data to use the service and help verify transactions. This makes the process slow or computationally impractical for some.

In a paper being presented at the Network and Distributed System Security Symposium next month, the MIT researchers introduce Vault, a cryptocurrency that lets users join the network by downloading only a fraction of the total transaction data. It also incorporates techniques that delete empty accounts that take up space, and enables verifications using only the most recent transaction data that are divided and shared across the network, minimizing an individual user’s data storage and processing requirements.

In experiments, Vault reduced the bandwidth for joining its network by 99 percent compared to Bitcoin and 90 percent compared to Ethereum, which is considered one of today’s most efficient cryptocurrencies. Importantly, Vault still ensures that all nodes validate all transactions, providing tight security equal to its existing counterparts.

“Currently there are a lot of cryptocurrencies, but they’re hitting bottlenecks related to joining the system as a new user and to storage. The broad goal here is to enable cryptocurrencies to scale well for more and more users,” says co-author Derek Leung, a graduate student in the Computer Science and Artificial Intelligence Laboratory (CSAIL).

Joining Leung on the paper are CSAIL researchers Yossi Gilad and Nickolai Zeldovich, who is also a professor in the Department of Electrical Engineering and Computer Science (EECS); and recent alumnus Adam Suhl ’18.

Vaulting over blocks

Each block in a cryptocurrency network contains a timestamp, its location in the blockchain, and fixed-length string of numbers and letters, called a “hash,” that’s basically the block’s identification. Each new block contains the hash of the previous block in the blockchain. Blocks in Vault also contain up to 10,000 transactions — or 10 megabytes of data — that must all be verified by users. The structure of the blockchain and, in particular, the chain of hashes, ensures that an adversary cannot hack the blocks without detection.

New users join cryptocurrency networks, or “bootstrap,” by downloading all past transaction data to ensure they’re secure and up to date. To join Bitcoin last year, for instance, a user would download 500,000 blocks totaling about 150 gigabytes. Users must also store all account balances to help verify new users and ensure users have enough funds to complete transactions. Storage requirements are becoming substantial, as Bitcoin expands beyond 22 million accounts.

The researchers built their system on top of a new cryptocurrency network called Algorand — invented by Silvio Micali, the Ford Professor of Engineering at MIT — that’s secure, decentralized, and more scalable than other cryptocurrencies.

With traditional cryptocurrencies, users compete to solve equations that validate blocks, with the first to solve the equations receiving funds. As the network scales, this slows down transaction processing times. Algorand uses a “proof-of-stake” concept to more efficiently verify blocks and better enable new users join. For every block, a representative verification “committee” is selected. Users with more money — or stake — in the network have higher probability of being selected. To join the network, users verify each certificate, not every transaction.

But each block holds some key information to validate the certificate immediately ahead of it, meaning new users must start with the first block in the chain, along with its certificate, and sequentially validate each one in order, which can be time-consuming. To speed things up, the researchers give each new certificate verification information based on a block a few hundred or 1,000 blocks behind it — called a “breadcrumb.” When a new user joins, they match the breadcrumb of an early block to a breadcrumb 1,000 blocks ahead. That breadcrumb can be matched to another breadcrumb 1,000 blocks ahead, and so on.

“The paper title is a pun,” Leung says. “A vault is a place where you can store money, but the blockchain also lets you ‘vault’ over blocks when joining a network. When I’m bootstrapping, I only need a block from way in the past to verify a block way in the future. I can skip over all blocks in between, which saves us a lot of bandwidth.”

Divide and discard

To reduce data storage requirements, the researchers designed Vault with a novel “sharding” scheme. The technique divides transaction data into smaller portions — or shards — that it shares across the network, so individual users only have to process small amounts of data to verify transactions.

To implement sharing in a secure way, Vault uses a well-known data structure called a binary Merkle tree. In binary trees, a single top node branches off into two “children” nodes, and those two nodes each break into two children nodes, and so on.

In Merkle trees, the top node contains a single hash, called a root hash. But the tree is constructed from the bottom, up. The tree combines each pair of children hashes along the bottom to form their parent hash. It repeats that process up the tree, assigning a parent node from each pair of children nodes, until it combines everything into the root hash. In cryptocurrencies, the top node contains a hash of a single block. Each bottom node contains a hash that signifies the balance information about one account involved in one transaction in the block. The balance hash and block hash are tied together.

To verify any one transaction, the network combines the two children nodes to get the parent node hash. It repeats that process working up the tree. If the final combined hash matches the root hash of the block, the transaction can be verified. But with traditional cryptocurrencies, users must store the entire tree structure.

With Vault, the researchers divide the Merkle tree into separate shards assigned to separate groups of users. Each user account only ever stores the balances of the accounts in its assigned shard, as well as root hashes. The trick is having all users store one layer of nodes that cuts across the entire Merkle tree. When a user needs to verify a transaction from outside of their shard, they trace a path to that common layer. From that common layer, they can determine the balance of the account outside their shard, and continue validation normally.

“Each shard of the network is responsible for storing a smaller slice of a big data structure, but this small slice allows users to verify transactions from all other parts of network,” Leung says.

Additionally, the researchers designed a novel scheme that recognizes and discards from a user’s assigned shard accounts that have had zero balances for a certain length of time. Other cryptocurrencies keep all empty accounts, which increase data storage requirements while serving no real purpose, as they don’t need verification. When users store account data in Vault, they ignore those old, empty accounts.


Source
Author: Rob Matheson
Image Credit

MIT Scientists Score Breakthrough in Finding Bacteria That Produce Electricity

Deeps in mines, at the bottom of lakes, and even in your own gut, bacteria are hard at work producing electricity in order to survive in environments low in oxygen.

These potent little power producers have been used in speculative experiments and one day may power everything from batteries to “biohomes“.

There are many types of bacteria capable of producing electricity, but some are better at it than others. The trouble with these bacteria is that they are difficult and expensive to grow in a lab setting, slowing down our ability to develop new technologies with them.

A new technique developed by MIT engineers makes sorting and identifying electricity-producing bacteria easier than ever before which may make them more readily available for us in technological applications.

Electricity-producing bacteria are able to pull off the trick by producing electrons within their cells and releasing them through tiny channels in their cell membranes in a process called extracellular electron transfer, or EET.

Current processes for identifying the electricity producing capabilities of bacteria involved measuring the activity of EET proteins but this is a daunting and time consuming process.

Researchers sometimes use a process called dielectrophoresis to separate two kinds of bacteria based on their electrical properties. They can use this process to differentiate between two different kinds of cells, such as cells from a frog and cells from a bird.

But the MIT team’s study separated cells based on a much more minute difference, their ability to produce electricity.

By applying small voltages to bacteria strains in an hourglass-shaped microfluidic channel the team was able to separate and measure the different kinds of closely related cells.

By noting the voltage required to manipulate bacteria and recording the cell’s size researchers were able to calculate each bacteria’s polarizability – how easy it is for a cell to produce electricity in an electric field.

Their study concluded that bacteria with a higher polarizability were also more active electricity producers.

Next the team will begin testing bacteria already thought to be strong candidates for future power production.

If their observations on polarizability hold true for these other bacteria, this new technique could make electricity-producing bacteria more accessible than ever before.


Source
Author: JACOB BANAS
Image Credit

Breaking: Harvard, Stanford, & MIT Have All Invested in Cryptocurrency Funds

At least five more university endowments have invested in cryptocurrency funds, suggesting that the “herd” of institutional investors is finally beginning to place at least a small bet on the nascent asset class.

As first reported by The Information, a cadre of major educational institutions including Harvard University, Stanford University, Massachusetts Institute of Technology, Dartmouth College, and the University of North Carolina have each invested in at least one cryptocurrency fund through their respective endowments.

Citing an unnamed source familiar with the investments, the publication reported that these five university endowments have invested tens of millions of dollars in these funds, which in turn invest in both physical cryptocurrencies and equity in cryptocurrency companies.

CCN previously reported that Yale University, which controls the second-largest university endowment next to Harvard, had allocated a portion of its $29.4 billion in assets into two cryptocurrency funds operated by Andreessen Horowitz (a16z) and Paradigm.

Even with these investments, the six universities that are now said to have invested in crypto funds still have very little exposure to this asset class. Nevertheless, the fact that they are engaging with the market at all could help legitimize the space.

As The Information journalist Jon Victor explained:

“A move by endowments into funds that will directly bet on cryptocurrencies signals a major shift in investor sentiment toward the asset class, in the same way that institutions over the past decade became more willing to invest in private tech companies. Backing from such closely watched institutions could help validate cryptocurrencies, which are still considered too risky by many institutional investors.”

Cryptocurrency investors and analysts such as Mike Novogratz had long predicted that a “herd” of institutional investors would power the next bitcoin bull market. Ari Paul, a cryptocurrency fund manager and a former portfolio manager at the University of Chicago’s endowment, said in April that he believed that a number of institutions were interested in investing in cryptocurrency but were waiting for major names such as Yale to make the first move so that they would have an “excuse” to do so themselves.

Notably, though institutional investors are generally viewed as having a more sober view of cryptoassets than retail investors, a recent survey by Wall Street strategy firm Fundstrat found that institutions that have already invested in cryptocurrency are actually more optimistic about bitcoin’s near-term prospects than retail investors.


Source
Author: Josiah Wilmoth
Image Credit

MIT Is Testing A Smart Contract-Powered Bitcoin Lightning Network

An MIT test is providing a rare glimpse of how bitcoin might truly work at scale.

Revealed last week, the prestigious U.S. university has been quietly demoing an experimental use case for bitcoin’s lightning network, one that showcases how it might be combined with smart contracts to not only handle millions of transactions, but do so with a greater degree of complexity.


Join in the fun and play on the world’s First Hybrid on-line Casino with BTC and Fiat currency payments. Check on-line for latest promotions


Modeled within the school’s Digital Currency Initiative, started in 2015 as a way to further R&D on cryptocurrencies, the test envisions a system wherein transactions would take place automatically in the case of defined external events, based on say today’s weather or the current price of U.S. dollars.

This is possible due to MIT’s creative use of so-called “oracles,” trusted entities meant to broadcast data to smart contracts. For this demo, researchers Tadge Dryja and Alin S. Dragos built a test oracle to broadcast the recent price of U.S. dollars in satoshis, the smallest unit of bitcoins, which anyone can grab and use for their smart contracts.

It’s a notable step forward for the idea, one first proposed by lightning inventor Dryja last summer. However, this is the first time it’s been implemented as a prototype with working code.

Dragos told CoinDesk:

“We built this as a standalone feature of our lightning network software. We chose data what we thought would be cool, U.S. dollars, but it could be any data you want, whether weather or a stock.”

Dragos stressed that the demo is “experimental” and “shouldn’t be used for real money.” That said, he and other MIT researchers are convinced that with the help of the lightning network, bitcoin might one day scale to capacities originally envisioned by its early users.

As part of that work, MIT researchers have already created an implementation for the lightning network called lit, and this oracle code is an add-on of that work.

“We at DCI, we really believe in the lightning network,” Dragos said. “Bitcoin doesn’t scale very well. I decided there has to be something better. Turns out what’s better is lightning. It’s the way to scale.”

Bitcoin smart contracts

But while lightning provides scale, smart contracts add other new functionality to bitcoin. For example, should the tech in MIT’s test be implemented, you could make some sort of a bet based on what’s happening in the world.

Or, in this case, a futures contract. Alice promises to pay Bob whatever the price of dollars is in satoshis on a certain day, say Friday. If a dollar is worth 12,150 satoshis by the end of the week, then she will end up paying that.

It’s a kind of advanced smart contract use case that is usually not associated with bitcoin.

“When folks think smart contracts, they think ethereum. Their scripting language is much richer,” Dragos admitted.

But, he argues that with some workarounds, bitcoin can do the same thing.

“It’s not as developer friendly because bitcoin didn’t go in that direction, but you can use it. You have to be a little creative,” Dragos said.

In short, it uses Dryja’s “discreet log contracts” scheme to broadcast data to the smart contracts. One of the most important advantages of this scheme is scalability, because most of the data doesn’t need to be stored on the bitcoin blockchain.

The other is privacy, since oracles don’t have any way of knowing who’s using the data they’re broadcasting.

“We’re introducing a model where oracles are not aware of who’s using the data they’re using,” Dragos said.

Some ‘quandaries’

But while this simple demo is now complete, Dragos and Dryja think there are many outstanding questions and “quandaries,” as Dragos put it. “From the individual oracle’s perspective, they’re going to want to make some money. We’re going to have to understand that,” Dragos said.

Another is that the oracle at this point is trusted. But there might be a way to minimize this trust by allowing a user to use many oracles at once.

But there’s a certain point where MIT DCI hopes to stop working on the technology and pass it off to someone else.

“We’re working with companies that might implement this,” Dragos said. And though he couldn’t name names, he mentioned they are “big company” partners of the DCI.

The hope is these bigger companies will be better at understanding what normal users want from the software. So, while MIT DCI built a prototype demonstrating how the underlying technology really works, they haven’t produced an app as mindlessly easy to use as say, Venmo or Facebook.

“UX is not our core expertise,” Dragos said.

Now it’s open for people to use for whatever oracle data they want. So, it’s up to the community to decide if it’s worthwhile to use or not.

“It’s a hard guess. It could be a significant deal if people use it. But we don’t know what people are going to be using it for,” he added.

Dragos stated:

“New technologies are available all the time, that doesn’t mean they end up making it though.”


Here at Dollar Destruction, we endeavor to bring to you the latest, most important news from around the globe. We scan the web looking for the most valuable content and dish it right up for you! The content of this article was provided by the source referenced. Dollar Destruction does not endorse and is not responsible for or liable for any content, accuracy, quality, advertising, products or other materials on this page. As always, we encourage you to perform your own research!

Source
Author: Alyssa Hertig
Image Credit