Review
Computer Science, Information Systems
Dimitrios M. Thilikos
Summary: Compactor is introduced as a general data-reduction concept for parametrized counting problems. It consists of a condenser and an extractor, attempting to formalize the notion of preprocessing for counting problems.
COMPUTER SCIENCE REVIEW
(2021)
Article
Computer Science, Artificial Intelligence
Yeonghun Kang, Hyunsoo Park, Berend Smit, Jihan Kim
Summary: Metal-organic frameworks (MOFs) are crystalline porous materials with tunable building blocks. Machine learning approach can explore the vast chemical space of MOFs by predicting their properties. MOFTransformer, a pre-trained multi-modal transformer, achieves state-of-the-art results for property prediction and provides chemical insights through feature analysis.
NATURE MACHINE INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Gido M. van de Ven, Tinne Tuytelaars, Andreas S. Tolias
Summary: Deep neural networks face challenges in continual learning, with different scenarios of continual learning having varying challenges and effectiveness. Distinguishing between task-incremental, domain-incremental, and class-incremental learning is an important foundation for organizing the continual learning field.
NATURE MACHINE INTELLIGENCE
(2022)
Article
Automation & Control Systems
Rafael Massambone, Eduardo Fontoura Costa, Elias Salomao Helou
Summary: In this article, a stochastic incremental subgradient algorithm is proposed for minimizing a sum of convex functions. The algorithm uses partial subgradients sequentially, and the sequence of subgradients is determined by a general Markov chain. The algorithm is suitable for networks due to its ability to handle stochastic information flow. The convergence of the algorithm to a weighted objective function is proven, where the weights are determined by the Cesaro limiting probability distribution of the Markov chain. The Cesaro limiting distribution allows for more flexibility in handling general weighted objective functions compared to previous works.
IEEE TRANSACTIONS ON AUTOMATIC CONTROL
(2023)
Article
Engineering, Electrical & Electronic
Ben Adcock, Simone Brugiapaglia, Matthew King-Roskamp
Summary: The study focuses on the application of the sparse in levels model in compressive imaging, and proposes new stable and robust uniform recovery guarantees, expanding the current research scope available under standard sparsity.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2021)
Article
Physics, Multidisciplinary
Sergey Bravyi, Anirban Chowdhury, David Gosset, Pawel Wocjan
Summary: This article studies the computational complexity of approximating the physical properties of a quantum many-body system in thermal equilibrium. A classical algorithm is proposed with polynomial runtime to approximate the free energy of a given 2-local Hamiltonian. The equivalence between approximating the free energy of local Hamiltonians and other natural tasks in condensed-matter physics and quantum computing is established, suggesting that the simulation of quantum many-body systems may capture the complexity of various computational problems. Finally, state-of-the-art classical and quantum algorithms for approximating the free energy are summarized and improvements are proposed.
Article
Physics, Multidisciplinary
Ewin Tang
Summary: The study introduces a new classical algorithm input model that captures the features and nuances of quantum linear algebra algorithms. Through this model, the authors describe classical analogs to quantum algorithms for principal component analysis and nearest-centroid clustering.
PHYSICAL REVIEW LETTERS
(2021)
Article
Multidisciplinary Sciences
Axin Fan, Tingfa Xu, Geer Teng, Xi Wang, Yuhan Zhang, Chang Xu, Xin Xu, Jianan Li
Summary: Polarization multispectral imaging (PMI) is widely used to characterize the physicochemical properties of objects. However, traditional PMI methods are time-consuming and require extensive storage resources. This paper presents a publicly available database of full-Stokes polarization multispectral images (FSPMI), measured using an established system, which may greatly facilitate PMI development and application.
Article
Engineering, Electrical & Electronic
Yash Sherry, Neil C. Thompson
Summary: Algorithms are crucial in computer science, enabling scientists to tackle larger problems and explore new domains. However, there is uncertainty regarding the extent and generalizability of progress in algorithms.
PROCEEDINGS OF THE IEEE
(2021)
Article
Engineering, Civil
Behrang Assemi, Alexander Paz, Douglas Baker
Summary: The study introduced a metaheuristic optimization algorithm to integrate bay-level parking occupancy snapshots and payment data, resulting in a highly accurate parking occupancy estimation model. The algorithm showed 76% accuracy in data integration, while the best model achieved an R-2 above 94% and a low RMSE of 1.2 for occupied bays when tested with integrated data samples.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Editorial Material
Multidisciplinary Sciences
Floris van Breugel, Bingni W. Brunton
Summary: By using genetic manipulations, scientists have cleverly applied perception-altering technology to gain insights into how fruit flies localize the source of smells by following tendrils of airborne odor plumes.
Article
Multidisciplinary Sciences
Elyse A. Schriber, Daniel W. Paley, Robert Bolotovsky, Daniel J. Rosenberg, Raymond G. Sierra, Andrew Aquila, Derek Mendez, Frederic Poitevin, Johannes P. Blaschke, Asmit Bhowmick, Ryan P. Kelly, Mark Hunter, Brandon Hayes, Derek C. Popple, Matthew Yeung, Carina Pareja-Rivera, Stella Lisova, Kensuke Tono, Michihiro Sugahara, Shigeki Owada, Tevye Kuykendall, Kaiyuan Yao, P. James Schuck, Diego Solis-Ibarra, Nicholas K. Sauter, Aaron S. Brewster, J. Nathan Hohman
Summary: The increase in inorganic-organic hybrid materials has led to a bottleneck in their characterization, which can be addressed using small-molecule serial femtosecond X-ray crystallography as a solution for determining their crystal structures.
Article
Physics, Multidisciplinary
Akel Hashim, Ravi K. Naik, Alexis Morvan, Jean-Loup Ville, Bradley Mitchell, John Mark Kreikebaum, Marc Davis, Ethan Smith, Costin Iancu, Kevin P. O'Brien, Ian Hincks, Joel J. Wallman, Joseph Emerson, Irfan Siddiqi
Summary: The successful implementation of algorithms on quantum processors requires accurate control of quantum bits, but coherent errors can severely limit performance. Randomized compiling can convert coherent errors into stochastic noise, reducing unpredictable errors and enabling accurate prediction of algorithmic performance. This approach demonstrates significant performance gains and accurately predicts algorithm performance on modern-day noisy quantum processors, paving the way for scalable quantum computing.
Article
Multidisciplinary Sciences
Shivam Garg, Kirankumar Shiragur, Deborah M. Gordon, Moses Charikar
Summary: Colonies of arboreal turtle ants create networks of trails in the canopy of tropical forests, using volatile pheromones and a decision rule based on pheromone levels. These networks approximately minimize the number of vertices and can solve the shortest path problem without central control or significant computational resources. A biologically plausible model based on reinforced random walk on a graph is proposed to explain this phenomena and provides algorithms for solving shortest path problems. Flow rate and decision rules based on pheromone level division are crucial for convergence to shortest paths.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2023)
Article
Automation & Control Systems
Xi Wang, Zhipeng Tu, Yiguang Hong, Yingyi Wu, Guodong Shi
Summary: This paper proposes online gradient descent and online bandit algorithms over Riemannian manifolds for full information and bandit feedback settings. The performance of these algorithms is evaluated by establishing a series of upper bounds on Hadamard manifolds, and the theoretical findings are validated through numerical studies.
JOURNAL OF MACHINE LEARNING RESEARCH
(2023)