Artificial Intelligence is a sphere of innovation that is now being deployed everywhere. QMware has identified a role in this field for quantum computing, and we are currently developing a market-ready quantum dialogue system with huge potential for application in AI.

 

Watch the movie about PlanQK: AI and Quantum Computing

Find more information here: Home – PlanQK

The Quantum Machine Learning Algorithms mimic intelligent behavior in an environment set by the rules of the Quantum Domain, such as the machine tries to become best in a game. Processes in organizations can be understood as and divided into clearly defined little universes with certain natural laws induced by the restricting functions and parameters of the Quantum Domain. A Quantum mechanical operator then produces a set of Eigenvalues in which the answer of the system is optimized according a certain task, which marks the solution of the given problem.

In this unique manner organizations can project their processes into the Quantum Machine Learning Algorithms and optimize the throughput inn real time for the most administrative challenges. If it comes to NP-hard problems which scale their number of calculation steps exponentially with the cardinality of their input sets, we can expect not to solve the practical problems in real time anymore, but upcoming native Quantum Computers could and will overcome these hindrances in certain cases.


Quantum Information Processing

QMware develops state-of-the-art quantum approaches that are capable of solving industry-important problems efficiently than conventional methods. Our expertise lies at the intersection of industrys most formidable cases and the high-end quantum and classical software development. We are extensively working on various optimization frameworks using quantum annealers and universal quantum computational networks. Cutting edge hybrid solutions developed by QMware will pave the way to a new level of efficiency in information processing techniques and deep learning.


Quantum Advantage

While the quantum computing community is at the beginning of embracing the capabilities of quantum devices, we already know that the quantum information processing will be a revolution that the world has not yet seen. Today as quantum computers are at an early stage, our developments are focused on leveraging the optimal use of quantum resources by combining the high-performance classical approaches and quantum algorithms in a hybrid solution.

Our team implemented a hybrid quantum algorithm for solving one of the emergent pressing problems in physical simulations and deep learning, linear systems of equations, with exponential speedup that utilizes quantum phase estimation, one of the exemplary core protocols for quantum computing.

We experimentally solve a 130,000-dimensional problem on a 20-qubit superconducting device, which is a record in a linear system solution on quantum computers. In comparison with the most significant supercomputers in history, our solution shows a dramatic increase in speed, which is experimental evidence that modern quantum computing surpasses the classical computing industry that reigned in the 20th century.

As a result, the developed scalable algorithm shows supremacy over classical computers, demonstrates advantages of quantum data processing via phase estimation, and holds high promise for meeting practically relevant challenges. We envision that the planned large-scale implementation will stimulate major players in the quantum computing industry to demonstrate the hardware actually achieving the quantum supremacy.

At QMware, we believe that quantum technologies will massively assist the most advanced machine learning frameworks. Highly autonomous systems that outperform humans at most economically valuable work requires large computational resources, which often limit their performance. The hybrid classical-quantum solvers implemented in a quantum neural network are considerable faster to train and make predictions.

As an illustrative example, we develop quantum neural networks, the Quantum Boltzmann Machine, for image recognition and time series forecasting. We have implemented the machine using D-Wave quantum annealer and have shown that our quantum network successfully recognizes the handwritten numbers and predicts the number of students that enter the MIPT campus based on the previous data.

Such a machine can show quantum advantage due to its quantum nature and, therefore, the speed of prediction. With the improvement in the architecture of state-of-the-art quantum annealers developed model can not be realized with purely classical methods indicating the urgent need for the efficient quantum embeddings in conventional machine learning solutions.

Nowadays, machine learning plays an increasing role in everyday life. With new quantum technological capabilities, QMware connects quantum technologies and machine learning. At QMware, we develop new quantum machine learning tools that enhance machine learning with quantum effects and improve quantum technologies with artificial intelligence.

We explore the possibilities of speeding-up machine learning algorithms with quantum computing capabilities. At QMware, we follow a hybrid approach, in which reinforcement learning techniques guide a quantum machine learning algorithm.

QMware provides consulting services and helps clients to find use cases of quantum machine learning techniques.

QMware has leading experts in the quantum machine learning field. Our recent research in quantum machine learning is featured on a cover of the Advanced Quantum Technologies journal. In this research, we developed a computer vision system that learns to predict quantum systems‘ behavior to understand the quantum advantage.