Andrew Vlasic
Research Subjects
Andrew Vlasic has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
- Anh Hoang Pham
Faculty of Materials for Energy, Shimane University, Matsue 690-8504, Japan
3 shared publications Salvatore Certo
Deloitte Consulting, LLP, Chicago, United States
3 shared publicationsHunter Grant
Deloitte Consulting, LLP, Chicago, United States
2 shared publicationsAnh Pham
Deloitte Consulting LLP
1 shared publication- Arthur J. Parzygnat
Experimental Study Group Massachusetts Institute of Technology 02139 Cambridge Massachusetts USA
1 shared publication - Salvador Aguinaga
Deloitte & Touche LLP
1 shared publication
Latest Publications
- Open Access
Towards structure-preserving quantum encodings
Harnessing the potential computational advantage of quantum computers for machine learning tasks relies on the uploading of classical data onto quantum computers through what are commonly referred to as quantum encodings. The choice of such encodings may vary substantially from one task to another, and there exist only a few cases where structure has provided insight into their design and implementation, such as symmetry in geometric quantum learning. Here, we propose the perspective that category theory offers a natural mathematical framework for analyzing encodings that respect structure inherent in datasets and learning tasks. We illustrate this with pedagogical examples, which include geometric quantum machine learning, quantum metric learning, topological data analysis, and more. Moreover, our perspective provides a language in which to ask meaningful and mathematically precise questions for the design of quantum encodings and circuits for quantum machine learning tasks.
Open AccessTowards structure-preserving quantum encodings
Harnessing the potential computational advantage of quantum computers for machine learning tasks relies on the uploading of classical data onto quantum computers through what are commonly referred to as quantum encodings. The choice of such encodings may vary substantially from one task to another, and there exist only a few cases where structure has provided insight into their design and implementation, such as symmetry in geometric quantum learning. Here, we propose the perspective that category theory offers a natural mathematical framework for analyzing encodings that respect structure inherent in datasets and learning tasks. We illustrate this with pedagogical examples, which include geometric quantum machine learning, quantum metric learning, topological data analysis, and more. Moreover, our perspective provides a language in which to ask meaningful and mathematically precise questions for the design of quantum encodings and circuits for quantum machine learning tasks.
- 8 November 2024
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 81 (1) , 1-15
Feature selection through quantum annealing
- 8 November 2024
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 81 (1) , 1-15
Feature selection through quantum annealing
- Open Access
QuOp: A Quantum Operator Representation for Nodes
We derive an intuitive and novel method to represent nodes in a graph with special unitary operators, or quantum operators, which does not require parameter training and is competitive with classical methods on scoring similarity between nodes. This method opens up future possibilities to apply quantum algorithms for NLP or other applications that need to detect anomalies within a network structure. Specifically, this technique leverages the advantage of quantum computation, representing nodes in higher dimensional Hilbert spaces. To create the representations, the local topology around each node with a predetermined number of hops is calculated and the respective adjacency matrix is used to derive the Hamiltonian. While using the local topology of a node to derive a Hamiltonian is a natural extension of a graph into a quantum circuit, our method differs by not assuming the quantum operators in the representation a priori, but letting the adjacency matrix dictate the representation. As a consequence of this simplicity, the set of adjacency matrices of size $2^n \times 2^n$ generates a sub-vector space of the Lie algebra of the special unitary operators, $\mathfrak{su}(2^n)$. This sub-vector space in turn generates a subgroup of the Lie group of special unitary operators, $\mathrm{SU}(2^n)$. Applications of our quantum embedding method, in comparison with the classical algorithms GloVe (a natural language processing embedding method) and FastRP (a general graph embedding method, display superior performance in measuring similarity between nodes in graph structures.Open AccessQuOp: A Quantum Operator Representation for Nodes
We derive an intuitive and novel method to represent nodes in a graph with special unitary operators, or quantum operators, which does not require parameter training and is competitive with classical methods on scoring similarity between nodes. This method opens up future possibilities to apply quantum algorithms for NLP or other applications that need to detect anomalies within a network structure. Specifically, this technique leverages the advantage of quantum computation, representing nodes in higher dimensional Hilbert spaces. To create the representations, the local topology around each node with a predetermined number of hops is calculated and the respective adjacency matrix is used to derive the Hamiltonian. While using the local topology of a node to derive a Hamiltonian is a natural extension of a graph into a quantum circuit, our method differs by not assuming the quantum operators in the representation a priori, but letting the adjacency matrix dictate the representation. As a consequence of this simplicity, the set of adjacency matrices of size $2^n \times 2^n$ generates a sub-vector space of the Lie algebra of the special unitary operators, $\mathfrak{su}(2^n)$. This sub-vector space in turn generates a subgroup of the Lie group of special unitary operators, $\mathrm{SU}(2^n)$. Applications of our quantum embedding method, in comparison with the classical algorithms GloVe (a natural language processing embedding method) and FastRP (a general graph embedding method, display superior performance in measuring similarity between nodes in graph structures. - Open Access
Complement Grover's Search Algorithm: An Amplitude Suppression Implementation
Grover's search algorithm was a groundbreaking advancement in quantum algorithms, displaying a quadratic speed-up of querying for items. Since the creation of this algorithm it has been utilized in various ways, including in preparing specific states for the general circuit. However, as the number of desired items increases so does the gate complexity of the sub-process that conducts the query. To counter this complexity, an extension of Grover's search algorithm is derived where the focus of the query is on the undesirable items in order to suppress the amplitude of the queried items. To display the efficacy the algorithm is implemented as a sub-process into QAOA and applied to a traveling salesman problem. For a basis of comparison, the results are compared against QAOA.
Open AccessComplement Grover's Search Algorithm: An Amplitude Suppression Implementation
Grover's search algorithm was a groundbreaking advancement in quantum algorithms, displaying a quadratic speed-up of querying for items. Since the creation of this algorithm it has been utilized in various ways, including in preparing specific states for the general circuit. However, as the number of desired items increases so does the gate complexity of the sub-process that conducts the query. To counter this complexity, an extension of Grover's search algorithm is derived where the focus of the query is on the undesirable items in order to suppress the amplitude of the queried items. To display the efficacy the algorithm is implemented as a sub-process into QAOA and applied to a traveling salesman problem. For a basis of comparison, the results are compared against QAOA.
- Open Access
Understanding the Mapping of Encode Data Through An Implementation of Quantum Topological Analysis
A potential advantage of quantum machine learning stems from the ability of encoding classical data into high dimensional complex Hilbert space using quantum circuits. Recent studies exhibit that not all encoding methods are the same when representing classical data since certain parameterized circuit structures are more expressive than the others. In this study, we show the difference in encoding techniques can be visualized by investigating the topology of the data embedded in complex Hilbert space. The technique for visualization is a hybrid quantum based topological analysis which uses simple diagonalization of the boundary operators to compute the persistent Betti numbers and the persistent homology graph. To augment the computation of Betti numbers within a NISQ framework, we suggest a simple hybrid algorithm. Through a illuminating example of a synthetic data set and the methods of angle encoding, amplitude encoding, and IQP encoding, we reveal topological differences with the encoding methods, as well as the original data. Consequently, our results suggest the encoding method needs to be considered carefully within different quantum machine learning models since it can strongly affect downstream analysis like clustering or classification.
Open AccessUnderstanding the Mapping of Encode Data Through An Implementation of Quantum Topological Analysis
A potential advantage of quantum machine learning stems from the ability of encoding classical data into high dimensional complex Hilbert space using quantum circuits. Recent studies exhibit that not all encoding methods are the same when representing classical data since certain parameterized circuit structures are more expressive than the others. In this study, we show the difference in encoding techniques can be visualized by investigating the topology of the data embedded in complex Hilbert space. The technique for visualization is a hybrid quantum based topological analysis which uses simple diagonalization of the boundary operators to compute the persistent Betti numbers and the persistent homology graph. To augment the computation of Betti numbers within a NISQ framework, we suggest a simple hybrid algorithm. Through a illuminating example of a synthetic data set and the methods of angle encoding, amplitude encoding, and IQP encoding, we reveal topological differences with the encoding methods, as well as the original data. Consequently, our results suggest the encoding method needs to be considered carefully within different quantum machine learning models since it can strongly affect downstream analysis like clustering or classification.
- Open Access
An Advantage Using Feature Selection with a Quantum Annealer
Feature selection is a technique in statistical prediction modeling that identifies features in a record with a strong statistical connection to the target variable. Excluding features with a weak statistical connection to the target variable in training not only drops the dimension of the data, which decreases the time complexity of the algorithm, it also decreases noise within the data which assists in avoiding overfitting. In all, feature selection assists in training a robust statistical model that performs well and is stable. Given the lack of scalability in classical computation, current techniques only consider the predictive power of the feature and not redundancy between the features themselves. Recent advancements in feature selection that leverages quantum annealing (QA) gives a scalable technique that aims to maximize the predictive power of the features while minimizing redundancy. As a consequence, it is expected that this algorithm would assist in the bias/variance trade-off yielding better features for training a statistical model. This paper tests this intuition against classical methods by utilizing open-source data sets and evaluate the efficacy of each trained statistical model well-known prediction algorithms. The numerical results display an advantage utilizing the features selected from the algorithm that leveraged QA.
Open AccessAn Advantage Using Feature Selection with a Quantum Annealer
Feature selection is a technique in statistical prediction modeling that identifies features in a record with a strong statistical connection to the target variable. Excluding features with a weak statistical connection to the target variable in training not only drops the dimension of the data, which decreases the time complexity of the algorithm, it also decreases noise within the data which assists in avoiding overfitting. In all, feature selection assists in training a robust statistical model that performs well and is stable. Given the lack of scalability in classical computation, current techniques only consider the predictive power of the feature and not redundancy between the features themselves. Recent advancements in feature selection that leverages quantum annealing (QA) gives a scalable technique that aims to maximize the predictive power of the features while minimizing redundancy. As a consequence, it is expected that this algorithm would assist in the bias/variance trade-off yielding better features for training a statistical model. This paper tests this intuition against classical methods by utilizing open-source data sets and evaluate the efficacy of each trained statistical model well-known prediction algorithms. The numerical results display an advantage utilizing the features selected from the algorithm that leveraged QA.
- Open Access
- 7 December 2022
- Published by Springer Nature in Research Square
Understanding the Mapping of Encode Data Through An Implementation of Quantum Topological Analysis
Topological Data Analysis (TDA) is a well-established field derived to give insight into the geometric structure of real-world data. However, many methods in TDA are computationally intensive. The method that computes the respective Betti number has been shown to obtain a speed-up from translating the algorithm into a quantum circuit. The quantum circuit to calculate a particular Betti number requires a significant number of gates and, without a small record of data, is currently unable to be implemented on a NISQ-era processor. Given this NISQ-era restriction, a hybrid-method is proposed that calculates the Euclidean distance of the encoded data and computes the desired Betti number. This method is applied to a toy data set with different encoding techniques. The empirical results show the noise within the data is intensified with each encoding method as there is a clear change in the geometric structure of the original data, exhibiting information loss.
- 7 December 2022
- Published by Springer Nature in Research Square
Open AccessUnderstanding the Mapping of Encode Data Through An Implementation of Quantum Topological Analysis
Topological Data Analysis (TDA) is a well-established field derived to give insight into the geometric structure of real-world data. However, many methods in TDA are computationally intensive. The method that computes the respective Betti number has been shown to obtain a speed-up from translating the algorithm into a quantum circuit. The quantum circuit to calculate a particular Betti number requires a significant number of gates and, without a small record of data, is currently unable to be implemented on a NISQ-era processor. Given this NISQ-era restriction, a hybrid-method is proposed that calculates the Euclidean distance of the encoded data and computes the desired Betti number. This method is applied to a toy data set with different encoding techniques. The empirical results show the noise within the data is intensified with each encoding method as there is a clear change in the geometric structure of the original data, exhibiting information loss.