-
Splitting and Parallelizing of Quantum Convolutional Neural Networks for Learning Translationally Symmetric Data
Authors:
Koki Chinzei,
Quoc Hoan Tran,
Kazunori Maruyama,
Hirotaka Oshima,
Shintaro Sato
Abstract:
The quantum convolutional neural network (QCNN) is a promising quantum machine learning (QML) model that is expected to achieve quantum advantages in classically intractable problems. However, the QCNN requires a large number of measurements for data learning, limiting its practical applications in large-scale problems. To alleviate this requirement, we propose a novel architecture called split-pa…
▽ More
The quantum convolutional neural network (QCNN) is a promising quantum machine learning (QML) model that is expected to achieve quantum advantages in classically intractable problems. However, the QCNN requires a large number of measurements for data learning, limiting its practical applications in large-scale problems. To alleviate this requirement, we propose a novel architecture called split-parallelizing QCNN (sp-QCNN), which exploits the prior knowledge of quantum data to design an efficient model. This architecture draws inspiration from geometric quantum machine learning and targets translationally symmetric quantum data commonly encountered in physics and quantum computing science. By splitting the quantum circuit based on translational symmetry, the sp-QCNN can substantially parallelize the conventional QCNN without increasing the number of qubits and improve the measurement efficiency by an order of the number of qubits. To demonstrate its effectiveness, we apply the sp-QCNN to a quantum phase recognition task and show that it can achieve comparable classification accuracy to the conventional QCNN while considerably reducing the measurement resources required. Due to its high measurement efficiency, the sp-QCNN can mitigate statistical errors in estimating the gradient of the loss function, thereby accelerating the learning process. These results open up new possibilities for incorporating the prior data knowledge into the efficient design of QML models, leading to practical quantum advantages.
△ Less
Submitted 27 February, 2024; v1 submitted 12 June, 2023;
originally announced June 2023.
-
RecipeMind: Guiding Ingredient Choices from Food Pairing to Recipe Completion using Cascaded Set Transformer
Authors:
Mogan Gim,
Donghee Choi,
Kana Maruyama,
Jihun Choi,
Hajung Kim,
Donghyeon Park,
Jaewoo Kang
Abstract:
We propose a computational approach for recipe ideation, a downstream task that helps users select and gather ingredients for creating dishes. To perform this task, we developed RecipeMind, a food affinity score prediction model that quantifies the suitability of adding an ingredient to set of other ingredients. We constructed a large-scale dataset containing ingredient co-occurrence based scores…
▽ More
We propose a computational approach for recipe ideation, a downstream task that helps users select and gather ingredients for creating dishes. To perform this task, we developed RecipeMind, a food affinity score prediction model that quantifies the suitability of adding an ingredient to set of other ingredients. We constructed a large-scale dataset containing ingredient co-occurrence based scores to train and evaluate RecipeMind on food affinity score prediction. Deployed in recipe ideation, RecipeMind helps the user expand an initial set of ingredients by suggesting additional ingredients. Experiments and qualitative analysis show RecipeMind's potential in fulfilling its assistive role in cuisine domain.
△ Less
Submitted 14 October, 2022;
originally announced October 2022.
-
Timestamp Based Execution Control for C and Java Programs
Authors:
Kazutaka Maruyama,
Minoru Terada
Abstract:
Many programmers have had to deal with an overwritten variable resulting for example from an aliasing problem. The culprit is obviously the last write-access to that memory location before the manifestation of the bug. The usual technique for removing such bugs starts with the debugger by (1) finding the last write and (2) moving the control point of execution back to that time by re-executing t…
▽ More
Many programmers have had to deal with an overwritten variable resulting for example from an aliasing problem. The culprit is obviously the last write-access to that memory location before the manifestation of the bug. The usual technique for removing such bugs starts with the debugger by (1) finding the last write and (2) moving the control point of execution back to that time by re-executing the program from the beginning. We wish to automate this. Step (2) is easy if we can somehow mark the last write found in step (1) and control the execution-point to move it back to this time.
In this paper we propose a new concept, position, that is, a point in the program execution trace, as needed for step (2) above. The position enables debuggers to automate the control of program execution to support common debugging activities. We have implemented position in C by modifying GCC and in Java with a bytecode transformer. Measurements show that position can be provided with an acceptable amount of overhead.
△ Less
Submitted 17 September, 2003;
originally announced September 2003.
