Discover more from Arkinfo Notes
Jyothikamalesh, Quantum Researcher, on Quantum Machine Learning (QML)
A discussion about Quantum Machine Learning, an emerging field in Quantum Computing and the Future of Quantum Computing
Greeting my fellow geeks! Welcome to this special edition of “Talks@Arkinfo”.
In this instalment of Arkinfo Talks, we are delving into the quantum realm with none other than Jyothikamalesh, a Quantum Computing Researcher. Balancing his role as an SDE Intern at Wells Fargo and seamlessly collaborating with the wizards at IBM Research tram, Jyothi's journey unfolds as a quantum leap. Jyothi's achievements include leading the charge at Stanford's Code in Place program and making waves with his groundbreaking paper at the International Springer Nature Group Conference.
In this exclusive tête-à-tête, Jyothikamalesh explains the concept of Quantum Machine Learning (QML) while always providing great insights on how to break into this industry.
Now, if you're as intrigued as I am, do yourself a favour: Follow Jyothikamalesh on LinkedIn. This is your golden ticket to the front row of quantum innovations and insights that are shaping our future. Don't just read about quantum leaps; be part of one!
Stay curious, stay quantum, and as always, keep exploring the unexplored with Arkinfo Notes! Let’s get started.
Want more such interviews delivered straight to your inbox for FREE? Subscribe now.
Can you briefly introduce yourself, and describe your journey from being an SDE Intern at Wells Fargo to becoming a Quantum Research Collaborator at IBM? What was the experience like working for each company?
Hello Arkinfo Readers,
I am Jyothikamalesh, currently a Intern analyst focusing on SDE/ML engineering at Wells Fargo, a position I earned through my victory in the Grand Academic Challenge conducted by Wells Fargo and T-AIM. Additionally, I am excited to share that I collaborated with IBM Quantum research team, after showcasing my passion and skills in quantum computing during a Quantum Machine Learning (QML) Hackathon organized by IIT-M and IBM.
In my capacity as an SDE at Wells Fargo, I have gained practical experience in the realm of computer science and financial technology. Simultaneously, my enthusiasm for quantum technologies led me to participate in the QML Hackathon, where my talent and dedication were recognized by IBM. This recognition led to my collaboration with IBM, where I explored the exciting world of quantum computing, contributing to research projects and gaining hands-on experience with IBM's advanced 27-qubit quantum computer.
These independent opportunities have allowed me to pursue my passion for both computer engineering and quantum physics. Balancing my responsibilities at Wells Fargo with my collaboration at IBM has been a fulfilling journey, providing me with valuable insights into two diverse yet interconnected fields. I am dedicated to making meaningful contributions to both sectors and continue to explore the endless possibilities at the intersection of technology and quantum physics.
What did the process of becoming a Section-Leader for the Code in Place program at Stanford look like? How did the experience help in shaping your interest in becoming a researcher.
Becoming a Section Leader for the Code in Place program at Stanford and my path toward research seemed mutually exclusive, yet both experiences proved nurturing in their own ways. Teaching diverse learners honed my inclusive communication skills, fostering empathy. This experience ran parallel to my research interests, emphasizing the significance of effective communication in both realms. This duality enhanced my adaptability, emphasizing the value of inclusive communication, a skill indispensable for both teaching and research.
Congratulations on getting your paper on Quantum Machine Learning accepted at the International Springer Nature Group Conference on Artificial Intelligence and Knowledge Processing for 2023. Can you briefly describe the thesis of your paper and the significance of its outcomes.
We developed a highly efficient Variational Quantum Eigensolver (VQE) approach tailored for the intricate Kagome lattice, a complex structure in condensed matter physics. By addressing the lattice's challenges, our method enables precise quantum simulations of its magnetic properties. The significance lies in bridging quantum computing with practical applications, enhancing our understanding of quantum magnetic phenomena. This work contributes to vital research areas like superconductivity and magnetism, advancing both Quantum Machine Learning and condensed matter physics. Acceptance at this esteemed conference underscores the practical impact of our research, marking a pivotal stride in quantum-powered material science exploration.
How did you choose the ansatz models for the VQE algorithm? What are the advantages and disadvantages of each model?
In our study, we explored various ansatz models provided as built-in modules, such as EfficientSU2 and Excitationpreserving ansatz. However, these predefined models didn't yield the desired simulation effects for the complex Kagome lattice. We encountered challenges related to converging to the global minimum due to the presence of numerous local minima, a common issue in variational algorithms. To overcome this, we tailored a hardware-efficient ansatz specifically suited for the Kagome lattice. This custom ansatz was designed to address the local minima problem, enabling us to obtain more accurate ground state estimates. The advantage of this approach was its tailored nature, ensuring a better fit for the intricacies of the Kagome lattice. However, a potential drawback could be its limited applicability to other lattice structures, making it essential to strike a balance between generality and specificity when designing ansatz models.
How did you measure the accuracy and efficiency of your VQE simulations? What are the sources of error and noise in your results?
To measure the accuracy and efficiency of our VQE simulations, we compared the relative errors between the classical simulations and quantum simulations of the Kagome lattice. The relative errors served as a metric to gauge the quality of our quantum approximation against classical computations. Regarding sources of error and noise, one of the main challenges in quantum computations is noise introduced due to imperfect physical implementations of quantum operations. This noise can arise from various sources, including decoherence, gate errors, and readout errors. Proper error mitigation techniques, such as error correction codes and noise-adaptive algorithms, were employed to mitigate these issues and enhance the reliability of our results.
What are the physical implications of your ground state estimates for the Kagome lattice? How do they relate to the phenomena of frustration and quantum magnetism?
Our accurate ground state estimates for the Kagome lattice have significant physical implications, particularly concerning the phenomena of frustration and quantum magnetism. The Kagome lattice is a prototypical frustrated quantum system where the geometry of the lattice leads to competing interactions between neighboring spins. By accurately simulating the ground state, we gain insights into the unique magnetic properties of this lattice. Frustration in the Kagome lattice results from the inability of the lattice's geometry to satisfy all the pairwise interactions simultaneously, leading to highly degenerate ground states. Quantum magnetism, on the other hand, refers to the quantum mechanical behavior of magnetic systems, and understanding it on the Kagome lattice provides valuable information about exotic quantum states and phase transitions. Our findings shed light on these phenomena, offering a deeper understanding of the interplay between frustration and quantum magnetism in complex quantum systems.
What are the challenges and limitations of using VQE for simulating complex quantum systems? How do you plan to overcome them in your future work?
One of the major challenges in using VQE for simulating complex quantum systems is the presence of Barren plateaus. Barren plateaus refer to the phenomenon where the gradients of the cost function with respect to the variational parameters become exponentially vanishing as the system size increases. This poses a significant hurdle, particularly in quantum machine learning research, as it leads to slow convergence and makes optimization intractable for large systems. In our study, a deeper understanding of the Hamiltonian allowed us to design a more efficient ansatz, addressing the Barren plateau issue to some extent. However, ongoing research is focused on developing novel techniques and algorithms to mitigate the impact of Barren plateaus further. Exploring techniques like adaptive gradient methods, quantum natural gradient optimization, and leveraging insights from quantum information theory are potential avenues for overcoming this challenge in our future work, enabling more robust and scalable simulations of complex quantum systems.
How would you describe the current trajectory of the Quantum Computing industry? What are some milestones we can expect in the next decade?
Quantum computing, akin to a helicopter, ascends with hardware and algorithm advancements, promising transformative applications. In the next decade, increased qubit count and enhanced stability will enable powerful simulations, paralleling helicopters accessing remote areas. Quantum supremacy, like helicopters surpassing conventional limits, will become prevalent. Rigorous error correction and fault tolerance ensure reliable quantum computations, akin to safety protocols. Quantum cloud services democratize access, fostering innovation across sectors, reminiscent of versatile car applications. Just as cars provide essential everyday transportation, classical computing remains vital for routine tasks, complementing quantum capabilities. This harmonious collaboration between quantum's groundbreaking potential and classical computing's reliability propels us into unprecedented realms of technological achievement, shaping our future endeavors.
What are your thoughts on “Quantum AI” - a merger of two groundbreaking technologies? What impact will the current progress in AI have on Quantum Computing?
The profound impact of quantum computing on Quantum internet, AI, NLP, and music composition is awe-inspiring. In Quantum internet, quantum entanglement promises unbreakable encryption, ensuring unprecedented data security. Quantum algorithms in AI and NLP accelerate complex computations, revolutionizing machine learning and natural language processing tasks. Additionally, the fusion of quantum computing and music composition leverages quantum particles' inherent randomness, enabling the creation of truly unique and innovative musical pieces. This generative ability, rooted in quantum randomness, unlocks endless creative potential, shaping a future where the boundaries between science and art blur, leading to groundbreaking discoveries and artistic expressions that redefine our understanding of the world.
Lastly, for readers fascinated by this conversation, what is one piece of advice or insight you'd like to share to motivate them to get started with the field?
Follow your curiosity! Stanford university’s Andrew Ng Advise to master Deep learning suits here too – dive into 20-50 research papers in the field and don't fear the learning curve. Skimming and understanding research papers is indeed a valuable skill that improves with practice; it's the great way to explore the mysteries of quantum science and Staying persistent has been really helpful in this journey.Happy to connect with fellow Innovation enthusiasts. Thanks for the Wonderful Interview, Quadri.