Tag Archives: play
#435474 Watch China’s New Hybrid AI Chip Power ...
When I lived in Beijing back in the 90s, a man walking his bike was nothing to look at. But today, I did a serious double-take at a video of a bike walking his man.
No kidding.
The bike itself looks overloaded but otherwise completely normal. Underneath its simplicity, however, is a hybrid computer chip that combines brain-inspired circuits with machine learning processes into a computing behemoth. Thanks to its smart chip, the bike self-balances as it gingerly rolls down a paved track before smoothly gaining speed into a jogging pace while navigating dexterously around obstacles. It can even respond to simple voice commands such as “speed up,” “left,” or “straight.”
Far from a circus trick, the bike is a real-world demo of the AI community’s latest attempt at fashioning specialized hardware to keep up with the challenges of machine learning algorithms. The Tianjic (天机*) chip isn’t just your standard neuromorphic chip. Rather, it has the architecture of a brain-like chip, but can also run deep learning algorithms—a match made in heaven that basically mashes together neuro-inspired hardware and software.
The study shows that China is readily nipping at the heels of Google, Facebook, NVIDIA, and other tech behemoths investing in developing new AI chip designs—hell, with billions in government investment it may have already had a head start. A sweeping AI plan from 2017 looks to catch up with the US on AI technology and application by 2020. By 2030, China’s aiming to be the global leader—and a champion for building general AI that matches humans in intellectual competence.
The country’s ambition is reflected in the team’s parting words.
“Our study is expected to stimulate AGI [artificial general intelligence] development by paving the way to more generalized hardware platforms,” said the authors, led by Dr. Luping Shi at Tsinghua University.
A Hardware Conundrum
Shi’s autonomous bike isn’t the first robotic two-wheeler. Back in 2015, the famed research nonprofit SRI International in Menlo Park, California teamed up with Yamaha to engineer MOTOBOT, a humanoid robot capable of driving a motorcycle. Powered by state-of-the-art robotic hardware and machine learning, MOTOBOT eventually raced MotoGPTM world champion Valentino Rossi in a nail-biting match-off.
However, the technological core of MOTOBOT and Shi’s bike vastly differ, and that difference reflects two pathways towards more powerful AI. One, exemplified by MOTOBOT, is software—developing brain-like algorithms with increasingly efficient architecture, efficacy, and speed. That sounds great, but deep neural nets demand so many computational resources that general-purpose chips can’t keep up.
As Shi told China Science Daily: “CPUs and other chips are driven by miniaturization technologies based on physics. Transistors might shrink to nanoscale-level in 10, 20 years. But what then?” As more transistors are squeezed onto these chips, efficient cooling becomes a limiting factor in computational speed. Tax them too much, and they melt.
For AI processes to continue, we need better hardware. An increasingly popular idea is to build neuromorphic chips, which resemble the brain from the ground up. IBM’s TrueNorth, for example, contains a massively parallel architecture nothing like the traditional Von Neumann structure of classic CPUs and GPUs. Similar to biological brains, TrueNorth’s memory is stored within “synapses” between physical “neurons” etched onto the chip, which dramatically cuts down on energy consumption.
But even these chips are limited. Because computation is tethered to hardware architecture, most chips resemble just one specific type of brain-inspired network called spiking neural networks (SNNs). Without doubt, neuromorphic chips are highly efficient setups with dynamics similar to biological networks. They also don’t play nicely with deep learning and other software-based AI.
Brain-AI Hybrid Core
Shi’s new Tianjic chip brought the two incompatibilities together onto a single piece of brainy hardware.
First was to bridge the deep learning and SNN divide. The two have very different computation philosophies and memory organizations, the team said. The biggest difference, however, is that artificial neural networks transform multidimensional data—image pixels, for example—into a single, continuous, multi-bit 0 and 1 stream. In contrast, neurons in SNNs activate using something called “binary spikes” that code for specific activation events in time.
Confused? Yeah, it’s hard to wrap my head around it too. That’s because SNNs act very similarly to our neural networks and nothing like computers. A particular neuron needs to generate an electrical signal (a “spike”) large enough to transfer down to the next one; little blips in signals don’t count. The way they transmit data also heavily depends on how they’re connected, or the network topology. The takeaway: SNNs work pretty differently than deep learning.
Shi’s team first recreated this firing quirk in the language of computers—0s and 1s—so that the coding mechanism would become compatible with deep learning algorithms. They then carefully aligned the step-by-step building blocks of the two models, which allowed them to tease out similarities into a common ground to further build on. “On the basis of this unified abstraction, we built a cross-paradigm neuron scheme,” they said.
In general, the design allowed both computational approaches to share the synapses, where neurons connect and store data, and the dendrites, the outgoing branches of the neurons. In contrast, the neuron body, where signals integrate, was left reconfigurable for each type of computation, as were the input branches. Each building block was combined into a single unified functional core (FCore), which acts like a deep learning/SNN converter depending on its specific setup. Translation: the chip can do both types of previously incompatible computation.
The Chip
Using nanoscale fabrication, the team arranged 156 FCores, containing roughly 40,000 neurons and 10 million synapses, onto a chip less than a fifth of an inch in length and width. Initial tests showcased the chip’s versatility, in that it can run both SNNs and deep learning algorithms such as the popular convolutional neural network (CNNs) often used in machine vision.
Compared to IBM TrueNorth, the density of Tianjic’s cores increased by 20 percent, speeding up performance ten times and increasing bandwidth at least 100-fold, the team said. When pitted against GPUs, the current hardware darling of machine learning, the chip increased processing throughput up to 100 times, while using just a sliver (1/10,000) of energy.
Although these stats are great, real-life performance is even better as a demo. Here’s where the authors gave their Tianjic brain a body. The team combined one chip with multiple specialized networks to process vision, balance, voice commands, and decision-making in real time. Object detection and target tracking, for example, relied on a deep neural net CNN, whereas voice commands and balance data were recognized using an SNN. The inputs were then integrated inside a neural state machine, which churned out decisions to downstream output modules—for example, controlling the handle bar to turn left.
Thanks to the chip’s brain-like architecture and bilingual ability, Tianjic “allowed all of the neural network models to operate in parallel and realized seamless communication across the models,” the team said. The result is an autonomous bike that rolls after its human, balances across speed bumps, avoids crashing into roadblocks, and answers to voice commands.
General AI?
“It’s a wonderful demonstration and quite impressive,” said the editorial team at Nature, which published the study on its cover last week.
However, they cautioned, when comparing Tianjic with state-of-the-art chips designed for a single problem toe-to-toe on that particular problem, Tianjic falls behind. But building these jack-of-all-trades hybrid chips is definitely worth the effort. Compared to today’s limited AI, what people really want is artificial general intelligence, which will require new architectures that aren’t designed to solve one particular problem.
Until people start to explore, innovate, and play around with different designs, it’s not clear how we can further progress in the pursuit of general AI. A self-driving bike might not be much to look at, but its hybrid brain is a pretty neat place to start.
*The name, in Chinese, means “heavenly machine,” “unknowable mystery of nature,” or “confidentiality.” Go figure.
Image Credit: Alexander Ryabintsev / Shutterstock.com Continue reading
Posted in Human Robots
#435260 How Tech Can Help Curb Emissions by ...
Trees are a low-tech, high-efficiency way to offset much of humankind’s negative impact on the climate. What’s even better, we have plenty of room for a lot more of them.
A new study conducted by researchers at Switzerland’s ETH-Zürich, published in Science, details how Earth could support almost an additional billion hectares of trees without the new forests pushing into existing urban or agricultural areas. Once the trees grow to maturity, they could store more than 200 billion metric tons of carbon.
Great news indeed, but it still leaves us with some huge unanswered questions. Where and how are we going to plant all the new trees? What kind of trees should we plant? How can we ensure that the new forests become a boon for people in those areas?
Answers to all of the above likely involve technology.
Math + Trees = Challenges
The ETH-Zürich research team combined Google Earth mapping software with a database of nearly 80,000 existing forests to create a predictive model for optimal planting locations. In total, 0.9 billion hectares of new, continuous forest could be planted. Once mature, the 500 billion new trees in these forests would be capable of storing about two-thirds of the carbon we have emitted since the industrial revolution.
Other researchers have noted that the study may overestimate how efficient trees are at storing carbon, as well as underestimate how much carbon humans have emitted over time. However, all seem to agree that new forests would offset much of our cumulative carbon emissions—still an impressive feat as the target of keeping global warming this century at under 1.5 degrees Celsius becomes harder and harder to reach.
Recently, there was a story about a Brazilian couple who replanted trees in the valley where they live. The couple planted about 2.7 million trees in two decades. Back-of-the-napkin math shows that they on average planted 370 trees a day, meaning planting 500 billion trees would take about 3.7 million years. While an over-simplification, the point is that planting trees by hand is not realistic. Even with a million people going at a rate of 370 trees a day, it would take 83 years. Current technologies are also not likely to be able to meet the challenge, especially in remote locations.
Tree-Bombing Drones
Technology can speed up the planting process, including a new generation of drones that take tree planting to the skies. Drone planting generally involves dropping biodegradable seed pods at a designated area. The pods dissolve over time, and the tree seeds grow in the earth below. DroneSeed is one example; its 55-pound drones can plant up to 800 seeds an hour. Another startup, Biocarbon Engineering, has used various techniques, including drones, to plant 38 different species of trees across three continents.
Drone planting has distinct advantages when it comes to planting in hard-to-access areas—one example is mangrove forests, which are disappearing rapidly, increasing the risk of floods and storm surges.
Challenges include increasing the range and speed of drone planting, and perhaps most importantly, the success rate, as automatic planting from a height is still likely to be less accurate when it comes to what depth the tree saplings are planted. However, drones are already showing impressive numbers for sapling survival rates.
AI, Sensors, and Eye-In-the-Sky
Planting the trees is the first step in a long road toward an actual forest. Companies are leveraging artificial intelligence and satellite imagery in a multitude of ways to increase protection and understanding of forested areas.
20tree.ai, a Portugal-based startup, uses AI to analyze satellite imagery and monitor the state of entire forests at a fraction of the cost of manual monitoring. The approach can lead to faster identification of threats like pest infestation and a better understanding of the state of forests.
AI can also play a pivotal role in protecting existing forest areas by predicting where deforestation is likely to occur.
Closer to the ground—and sometimes in it—new networks of sensors can provide detailed information about the state and needs of trees. One such project is Trace, where individual trees are equipped with a TreeTalker, an internet of things-based device that can provide real-time monitoring of the tree’s functions and well-being. The information can be used to, among other things, optimize the use of available resources, such as providing the exact amount of water a tree needs.
Budding Technologies Are Controversial
Trees are in many ways fauna’s marathon runners—slow-growing and sturdy, but still susceptible to sickness and pests. Many deforested areas are likely not as rich in nutrients as they once were, which could slow down reforestation. Much of the positive impact that said trees could have on carbon levels in the atmosphere is likely decades away.
Bioengineering, for example through CRISPR, could provide solutions, making trees more resistant and faster-growing. Such technologies are being explored in relation to Ghana’s at-risk cocoa trees. Other exponential technologies could also hold much future potential—for instance micro-robots to assist the dwindling number of bees with pollination.
These technologies remain mired in controversy, and perhaps rightfully so. Bioengineering’s massive potential is for many offset by the inherent risks of engineered plants out-competing existing fauna or growing beyond our control. Micro-robots for pollination may solve a problem, but don’t do much to address the root cause: that we seem to be disrupting and destroying integral parts of natural cycles.
Tech Not The Whole Answer
So, is it realistic to plant 500 billion new trees? The short answer would be that yes, it’s possible—with the help of technology.
However, there are many unanswered challenges. For example, many of areas identified by the ETH-Zürich research team are not readily available for reforestation. Some are currently reserved for grazing, others owned by private entities, and others again are located in remote areas or areas prone to political instability, beyond the reach of most replanting efforts.
If we do wish to plant 500 billion trees to offset some of the negative impacts we have had on the planet, we might well want to combine the best of exponential technology with reforestation as well as a move to other forms of agriculture.
Such an approach might also help address a major issue: that few of the proposed new forests will likely succeed without ensuring that people living in and around the areas where reforestation takes place become involved, and can reap rewards from turning arable land into forests.
Image Credit: Lillac/Shutterstock.com Continue reading
Posted in Human Robots