Hu Yoshida

Social Innovation Drives Computing Innovations for Powering Good

Blog Post created by Hu Yoshida Employee on Jun 7, 2019

“Social Innovation addresses the world’s social and environmental needs. It’s bigger than an individual or company. Social Innovation requires businesses and the entire society to work together toward a common goal. The goal of Hitachi’s Social Innovation can be summed 
up with two simple words: "Powering Good.”

 

We are in the beginnings of a new revolution that is described as the Cognitive revolution. What differentiates this from all previous revolutions, like the industrial revolution and the information revolution, is that it goes beyond the ability of technology to augment our physical capabilities to build things or to communicate things. The cognitive revolution is defined by technology’s ability to augment the cognitive potential of humans. This will be more disruptive to society than all previous revolutions.

                                                                                                                             

The cognitive revolution is viewed by Hitachi as an opportunity to further its corporate goal for Social Innovation which addresses the world’s social and environmental needs. The key technologies driving this revolution are cognitive technologies like AI, ML/DL, NLP, AR/VR, Video Analytics. AI, machine learning and deep learning will help to solve many societal problems, like climate change, crime, disease, and the challenges of enhancing the quality of life and standard of living in megacities.

 

These types of calculations are based on the calculating the probabilities of many possible choices - combinatorial optimization problems. The number of possibilities grow exponentially and outpaces the compute capabilities of today’s CPU’s and GPUs. The only way to address this is through the use of quantum computers.

 

Quantum computing takes advantage of the strange ability of subatomic particles to exist in more than one state at any time. I am not about to provide a tutorial here on quantum computing except to refer you to wikipedia. Suffice it to say that a classical, or regular computer, contains a processor which manipulates strings made up of bits. These bits can only have a value of either 0 or 1, depending on the electrical charge applied to them. A quantum computer replaces these binary bits with quantum bits - or qubits for short. These are quantum particles which contain multiple states existing simultaneously in superposition. Their information may be stored as the spin property of that particle, or its momentum, even location. A qubit is not limited like regular binary bits to two states. It can have many states, which gives the quantum computer its exponential processing power. A quantum computer comprised of 500 qubits would have a potential to do 2^500 calculations in a single step. This is an awesome number - 2^500 is infinitely more atoms than there are in the known universe.

This effect of superposition allows a qubit to perform many times more calculations at a time than a standard computer. It is ideal for combinatorial optimization problems. While quantum computers are suited for these types of problems, they are not replacements for standard computers which are better suited for transactional problems or playing Youtube videos.

 

In 2015, Google and NASA reported that their new 1097-qubit D-Wave quantum computer had solved an optimization problem in a few seconds. That’s 100 million times faster than a regular computer chip. They claimed that a problem their D-Wave 2X machine processed inside one second would take a classical computer 10,000 years to solve.

 

While quantum computers are available from a few vendors for test and development purposes. Many are dependent on super conducting materials which require temperatures in the near absolute zero temperature range to minimize the movement of electrons and atoms. The D-Wave quantum computer referenced above, is cooled to 15 millikelvin, which is approximately 180 times colder than interstellar space. It is not likely that you will see a D-Wave quantum computer in your data center soon even if you could afford it.

 

Hitachi has come up with an innovative way to solve combinatorial optimization problems, which are key to solving many societal issues. Hitachi’s innovation is the use of a complementary metal oxide semiconductor (CMOS) circuit which is similar to what we use in CPUs today. This computer can solve these problems without the need for a quantum computer and super conducting materials.The key to solving these problems is the use of a mathematical model called the Ising Model  which is used to research magnetic properties in the field of statistical mechanics. The Ising model consists of spins that can be arranged in a lattice and the spins can be oriented up or down. Each spin interacts with the spin of its neighbors creating energy H.

Solving combinatorial optimization problems usually consists of testing various combinations and searching for the best solution. Instead of testing every combination, the problem we want to solve is to first map it to the Ising model. By conducting a process called annealing we can converge the combination of spinning directions and thereby optimize the energy expenditure, H, of the Ising model. (Annealing is a method for removing distortions on the inside of iron and steel by slowly cooling it after heating it to high temperatures.) The computer that Hitachi Invented reproduces the convergence behavior through a CMOS circuit. The Ising model that was mapped with the combinatorial optimization problem, is converged to a state that expends a minimal amount of energy. This minimal state represents an optimal solution to the combinatorial problem without the need to test every combination.

 

This CMOS annealing computer with the Ising model is available in our research lab and is solving problems today. Engineers here in Santa Clara code the problems in python and upload them to the research lab cloud in Japan where the researchers do the mapping to the Ising model.  Researchers are working on developing a higher level language to make the model more accessible. The mapping language is similar to the use of machine coding before the advent of higher level languages in standard computers. The use of CMOS annealing will make it possible to deliver an Ising computer to solve combinatorial optimization problems at costs which would be similar to standard computers.

 

Meanwhile, a number of companies and research institutes around the world are working on “universal” quantum computers. These use quantum gates to handle quantum bits, or qubits, in more complex ways. This enables them to run more sophisticated algorithms than CMOS annealing or quantum annealing machines and deal with a broader range of applications. Hitachi continues to invest in quantum computer research at their Hitachi Cambridge Laboratory (“HCL”), working in collaboration with academic partners at the University of Cambridge and University College London. However, today, quantum computers need to maintain near-zero temperatures and remain free from magnetic interference, thermal noise, and mechanical vibration in order for qubits to maintain superposition—the dual states of both 0 and 1—which forms the basis of quantum calculations. This means that the availability of commercial quantum computers are still some years away.

 

There is a dark side to quantum computers. Its ability to factor large prime numbers allows them to break asymmetric encryption which is used by internet communications schemes like SSL and its blazing speed could brute force symmetric encryption which is used for encryption of data at rest. If we believe that quantum computers will be available in the next 10 years, we need to think about protecting the data that we encrypt today to ensure that it is protected beyond the time that quantum computers become available. While researchers are looking at how to prevent this with future quantum proof encryption schemes, NIST recommends that we increase the key sizes of symmetric encryption and hashing algorithms for data at rest to make it harder for brute force attacks. The current key size in use is 256, which represents 2^256 which is an astronomically high number of combinations, however a quantum computer with enough Qubits could blow through that in an instant. The National Security Agency (NSA) proposes a “Rule of Two” to double encrypt sensitive data with two keys that are generated completely independently.

 

In the meantime, Hitachi will be helping our customers solve many of the problems that are beyond the capabilities of standard computers with their innovative CMOS Annealing computers. These computers will be used for “powering good” helping to solve many societal issues.

 

 

Outcomes