-
A Novel Quantum Algorithm for Efficient Attractor Search in Gene Regulatory Networks
Authors:
Mirko Rossini,
Felix M. Weidner,
Joachim Ankerhold,
Hans A. Kestler
Abstract:
The description of gene interactions that constantly occur in the cellular environment is an extremely challenging task due to an immense number of degrees of freedom and incomplete knowledge about microscopic details. Hence, a coarse-grained and rather powerful modeling of such dynamics is provided by Boolean Networks (BNs). BNs are dynamical systems composed of Boolean agents and a record of the…
▽ More
The description of gene interactions that constantly occur in the cellular environment is an extremely challenging task due to an immense number of degrees of freedom and incomplete knowledge about microscopic details. Hence, a coarse-grained and rather powerful modeling of such dynamics is provided by Boolean Networks (BNs). BNs are dynamical systems composed of Boolean agents and a record of their possible interactions over time. Stable states in these systems are called attractors which are closely related to the cellular expression of biological phenotypes. Identifying the full set of attractors is, therefore, of substantial biological interest. However, for conventional high-performance computing, this problem is plagued by an exponential growth of the dynamic state space. Here, we demonstrate a novel quantum search algorithm inspired by Grover's algorithm to be implemented on quantum computing platforms. The algorithm performs an iterative suppression of states belonging to basins of previously discovered attractors from a uniform superposition, thus increasing the amplitudes of states in basins of yet unknown attractors. This approach guarantees that a new attractor state is measured with each iteration of the algorithm, an optimization not currently achieved by any other algorithm in the literature. Tests of its resistance to noise have also shown promising performance on devices from the current Noise Intermediate Scale Quantum Computing (NISQ) era.
△ Less
Submitted 16 August, 2024;
originally announced August 2024.
-
AutoNumerics-Zero: Automated Discovery of State-of-the-Art Mathematical Functions
Authors:
Esteban Real,
Yao Chen,
Mirko Rossini,
Connal de Souza,
Manav Garg,
Akhil Verghese,
Moritz Firsching,
Quoc V. Le,
Ekin Dogus Cubuk,
David H. Park
Abstract:
Computers calculate transcendental functions by approximating them through the composition of a few limited-precision instructions. For example, an exponential can be calculated with a Taylor series. These approximation methods were developed over the centuries by mathematicians, who emphasized the attainability of arbitrary precision. Computers, however, operate on few limited precision types, su…
▽ More
Computers calculate transcendental functions by approximating them through the composition of a few limited-precision instructions. For example, an exponential can be calculated with a Taylor series. These approximation methods were developed over the centuries by mathematicians, who emphasized the attainability of arbitrary precision. Computers, however, operate on few limited precision types, such as the popular float32. In this study, we show that when aiming for limited precision, existing approximation methods can be outperformed by programs automatically discovered from scratch by a simple evolutionary algorithm. In particular, over real numbers, our method can approximate the exponential function reaching orders of magnitude more precision for a given number of operations when compared to previous approaches. More practically, over float32 numbers and constrained to less than 1 ULP of error, the same method attains a speedup over baselines by generating code that triggers better XLA/LLVM compilation paths. In other words, in both cases, evolution searched a vast space of possible programs, without knowledge of mathematics, to discover previously unknown optimized approximations to high precision, for the first time. We also give evidence that these results extend beyond the exponential. The ubiquity of transcendental functions suggests that our method has the potential to reduce the cost of scientific computing applications.
△ Less
Submitted 13 December, 2023;
originally announced December 2023.
-
A deep scalable neural architecture for soil properties estimation from spectral information
Authors:
Flavio Piccoli,
Micol Rossini,
Roberto Colombo,
Raimondo Schettini,
Paolo Napoletano
Abstract:
In this paper we propose an adaptive deep neural architecture for the prediction of multiple soil characteristics from the analysis of hyperspectral signatures. The proposed method overcomes the limitations of previous methods in the state of art: (i) it allows to predict multiple soil variables at once; (ii) it permits to backtrace the spectral bands that most contribute to the estimation of a gi…
▽ More
In this paper we propose an adaptive deep neural architecture for the prediction of multiple soil characteristics from the analysis of hyperspectral signatures. The proposed method overcomes the limitations of previous methods in the state of art: (i) it allows to predict multiple soil variables at once; (ii) it permits to backtrace the spectral bands that most contribute to the estimation of a given variable; (iii) it is based on a flexible neural architecture capable of automatically adapting to the spectral library under analysis. The proposed architecture is experimented on LUCAS, a large laboratory dataset and on a dataset achieved by simulating PRISMA hyperspectral sensor. 'Results, compared with other state-of-the-art methods confirm the effectiveness of the proposed solution.
△ Less
Submitted 26 October, 2022;
originally announced October 2022.
-
On Location Relevance and Diversity in Human Mobility Data
Authors:
Maria Luisa Damiani,
Fatima Hachem,
Christian Quadri,
Matteo Rossini,
Sabrina Gaito
Abstract:
The theme of human mobility is transversal to multiple fields of study and applications, from ad-hoc networks to smart cities, from transportation planning to recommendation systems on social networks. Despite the considerable efforts made by a few scientific communities and the relevant results obtained so far, there are still many issues only partially solved, that ask for general and quantitati…
▽ More
The theme of human mobility is transversal to multiple fields of study and applications, from ad-hoc networks to smart cities, from transportation planning to recommendation systems on social networks. Despite the considerable efforts made by a few scientific communities and the relevant results obtained so far, there are still many issues only partially solved, that ask for general and quantitative methodologies to be addressed. A prominent aspect of scientific and practical relevance is how to characterize the mobility behavior of individuals. In this article, we look at the problem from a location-centric perspective: we investigate methods to extract, classify and quantify the symbolic locations specified in telco trajectories, and use such measures to feature user mobility. A major contribution is a novel trajectory summarization technique for the extraction of the locations of interest, i.e. attractive, from symbolic trajectories. The method is built on a density-based trajectory segmentation technique tailored to telco data, which is proven to be robust against noise. To inspect the nature of those locations, we combine the two dimensions of location attractiveness and frequency into a novel location taxonomy, which allows for a more accurate classification of the visited places. Another major contribution is the selection of suitable entropy-based metrics for the characterization of single trajectories, based on the diversity of the locations of interest. All these components are integrated in a framework utilized for the analysis of 100,000+ telco trajectories. The experiments show how the framework manages to dramatically reduce data complexity, provide high-quality information on the mobility behavior of people and finally succeed in grasping the nature of the locations visited by individuals.
△ Less
Submitted 20 October, 2020;
originally announced October 2020.
-
Learning Behavioral Representations of Human Mobility
Authors:
Maria Luisa Damiani,
Andrea Acquaviva,
Fatima Hachem,
Matteo Rossini
Abstract:
In this paper, we investigate the suitability of state-of-the-art representation learning methods to the analysis of behavioral similarity of moving individuals, based on CDR trajectories. The core of the contribution is a novel methodological framework, mob2vec, centered on the combined use of a recent symbolic trajectory segmentation method for the removal of noise, a novel trajectory generaliza…
▽ More
In this paper, we investigate the suitability of state-of-the-art representation learning methods to the analysis of behavioral similarity of moving individuals, based on CDR trajectories. The core of the contribution is a novel methodological framework, mob2vec, centered on the combined use of a recent symbolic trajectory segmentation method for the removal of noise, a novel trajectory generalization method incorporating behavioral information, and an unsupervised technique for the learning of vector representations from sequential data. Mob2vec is the result of an empirical study conducted on real CDR data through an extensive experimentation. As a result, it is shown that mob2vec generates vector representations of CDR trajectories in low dimensional spaces which preserve the similarity of the mobility behavior of individuals.
△ Less
Submitted 11 September, 2020; v1 submitted 10 September, 2020;
originally announced September 2020.
