Quantum computing is surpassing the current physical capabilities limited by classical computing, including that of supercomputers, promising significant transformations in various fields in just 5 to 10 years. According to Moore’s law, which marks advances in technology.
Fields such as optimization, chemistry, and medicine have been boosted by the many benefits of quantum computing. One of the most promising areas is Quantum Machine Learning (QML), which promises to provide us with faster, more powerful and more reliable algorithms.
One of these promises materializes in AWS, where you have access to QuEra–Aquila, which has 256 qubits based on neutral atoms. It may seem like a few qubits, however, consider the case in which 500 of these are avail able. The number of amplitudes for such a systemwould exceed the estimated number of atoms in the Universe.
The Evolution of Data Science into Quantum Data Science
However, today, hardly any people are talking about one of the roles that will revolutionize the job market when quantum computing is implemented in real environments. The new role is Quantum Data Scientist (QDS).
Currently, all companies are governed through data for informed deci sions, so it is not difficult to find at least one open position for Data Scientist (DS). However, DS has limitations in its data analyzes due to limited com putational capacity or poor or partial data collection.
The quantum algorithms reported to date promise quantum supremacy and improvement in the search of the hyperparameter space. Being able to report much more precise values, in less time.
A study [3], based on quantum Fourier regression for the optimization of hyperparameters in forecasts in the airline industry, demonstrates supremacy in terms of precision and speed of convergence. These results are achieved due to the stochastic nature and quantum computing capacity. For example, it is stated in case C that the hyperparameters were optimized over 1980 samples in 41 minutes 45 seconds with a R2test = 0.388. Running the algorithm they propose on 280 samples, it took 15 minutes 2 seconds to find the best hyperparameters with R2 = 0.415.
Another example of quantum capabilities is Grover’s algorithm that al lows you to find a record in a messy database in record time. With a complexity of √N you can report the desired record in a database of 1, 000, 000 records in a maximum of 1000 attempts. This is due to the use of the laws of quantum mechanics that allow interference, superposition, and entangle ment.
Quantum Machine Learning (QML) today
Currently, the role of QML is the translation of classical ML algorithms to quantum systems with the goal of demonstrating quantum supremacy. Many research groups work on the optimization of complex systems such as supply chains, finance, and logistics. Typically, the complexity of the problems is NP-Hard. Therefore, quantum computers can offer a better solution since they solve BQP type problems that are between P and PSPACE.
Quantum annealing is proven efficient in solving optimization problems (NP-hard) such as the Vehicle Routing Problem. Gonz´alez Bermejo, et al. [1], uses quantum processes to solve combinatorial optimization problems using entropy as an objective to force the exploration of potential solutions.
Distinguishing Quantum Data Science from QML
It is then worth asking: What differentiates QML from QDS? QML works on the creation and translation of traditional ML systems to quantum envi ronments.
The QDS has the focus of creating a methodology for quantum data processing, as well as its governance. The QDS must provide solutions that can be applied in real environments efficiently. It is expected that as this role advances in the business world and the current limitations of quantum computing are resolved, it will be applied to the entire phase of the data lifecycle. An approach oriented towards data analytics.
In addition, the QDS has a broader objective, since it must create systems and algorithms that allow everything from predictive analysis to analyzing Big Data. On the other hand, QML’s objective is limited to the implemen tation and improvement of ML algorithms.
Practical implications
ellow Underline (used with H2)
Quantum computing not only promises more processing power, efficiency, speed, and precision [2]; but also offers a new field of study for data scientists. To the extent that sensors that use the principles of quantum mechanics can be linked to quantum processors without having to go through a classical binary system, a world of possibilities opens up.
The data scientist will be able to work directly with nature data and will be able to obtain new records that are currently neither imaginable nor capturable. For example, with quantum gravity sensors, quantum interfer ometry could measure variations in gravitational fields that would be useful for geology.
Therefore, it will be essential for the QDS to create new data analysis methodologies that provide for concepts such as the interweaving of multidi mensional data.
The new skills paradigm for being a QDS
To prepare for new challenges, it is completely necessary that you master quantum mechanics. You must know the fundamental principles such as overlapping and entanglement.
In parallel, you must learn quantum programming. A QDS must under stand theory, but must also be able to apply it to data analysis. Currently, you have multiple Python frameworks on like Qiskit, PennyLane, etc.
Researchers from Arizona State University [4] have already written a paper at the IEEE 15th International Symposium on Autonomous Decen tralized System about how to learn QML.
So, in short, you need to be better than a software engineer at quantum mechanics and better at quantum programming than a quantum physicist. Knowing that the objective of the QDS is to perform data analytics.
Conclusions
In conclusion, the role of the future is that of the QDS. An expert at the intersection between business, quantum mechanics and artificial intelligence. The QDS is tasked with creating new lifecycle data use methodologies, as well as the governance and ethics of quantum mechanics-based data systems.
References
ellow Underline (used with H2)
[1] Sa´ul Gonz´alez-Bermejo, Guillermo Alonso-Linaje, and Parfait Atchade Adelomou. “GPS: A new TSP formulation for its generalizations type QUBO.” In: (Oct. 2021). url: http://arxiv.org/abs/2110.12158.
[2] Yazhen Wang. “When Quantum Computation Meets Data Science: Mak ing Data Science Quantum.” In: Harvard Data Science Review (Jan. 2022). doi: 10.1162/99608f92.ef5d8928.
[3] S. Consul-Pacareu et al. “Quantum Machine Learning hyperparameter search.” In: (Feb. 2023). url: http://arxiv.org/abs/2302.10298.
[4] Gennaro De Luca and Yinong Chen. “Teaching Quantum Machine Learn ing in Computer Science.” In: Institute of Electrical and Electronics En gineers Inc., 2023. isbn: 9781665464505. doi: 10.1109/ISADS56919. 2023.10092171.
Disclaimer: The author is completely responsible for the content of this article. The opinions expressed are their own and do not represent IEEE’s position nor that of the Computer Society nor its Leadership.
