Search

This work evaluates a forward-only learning algorithm on the MNIST dataset with hardware-in-the-loop training of a 4f optical correlator, achieving 87.6% accuracy with O(n2) complexity, compared to backpropagation, which achieves 88.8% accuracy with O(n2 log n) complexity.

While the human brain efficiently adapts to new tasks from a continuous stream of information, neural network models struggle to learn from sequential information without catastrophically forgetting previously learned tasks. This limitation presents a significant hurdle in deploying edge devices in real-world scenarios where information is presented in an inherently sequential manner. Active dendrites of pyramidal neurons play an important role in the brain ability to learn new tasks incrementally. By exploiting key properties of time-to-first-spike encoding and leveraging its high sparsity, we present a novel spiking neural network model enhanced with active dendrites. Our model can efficiently mitigate catastrophic forgetting in temporally-encoded SNNs, which we demonstrate with an end-of-training accuracy across tasks of 88.3% on the test set using the Split MNIST dataset. Furthermore, we provide a novel digital hardware architecture that paves the way for real-world deployment in edge devices. Using a Xilinx Zynq-7020 SoC FPGA, we demonstrate a 100-% match with our quantized software model, achieving an average inference time of 37.3 ms and an 80.0% accuracy., Comment: This work was accepted and presented at AICAS 2024

When devising recommendation services, it is important to account for the interests of all content providers, encompassing not only newcomers but also minority demographic groups. In various instances, certain provider groups find themselves underrepresented in the item catalog, a situation that can influence recommendation results. Hence, platform owners often seek to regulate the exposure of these provider groups in the recommended lists. In this paper, we propose a novel cost-sensitive approach designed to guarantee these target exposure levels in pairwise recommendation models. This approach quantifies, and consequently mitigate, the discrepancies between the volume of recommendations allocated to groups and their contribution in the item catalog, under the principle of equity. Our results show that this approach, while aligning groups exposure with their assigned levels, does not compromise to the original recommendation utility. Source code and pre-processed data can be retrieved at https://github.com/alessandraperniciano/meta-learning-strategy-fair-provider-exposure., Comment: Accepted at the 46th European Conference on Information Retrieval (ECIR 2024)

We present an algorithm to solve the CASSCF linear response equations that is both simple and efficient. The algorithm makes use of the well established symmetric and antisymmetric combinations of trial vectors, but further orthogonalizes them with respect to the scalar product induced by the response matrix. This leads to a standard, symmetric, block eigenvalue problem in the expansion subspace that can be solved by diagonalizing a symmetric, positive definite matrix half the size of the expansion space. Preliminary numerical tests show that the algorithm is robust and stable.

The standard photometric stereo model makes several assumptions that are rarely verified in experimental datasets. In particular, the observed object should behave as a Lambertian reflector and the light sources should be positioned at an infinite distance from it, along a known direction. Even when Lambert's law is approximately fulfilled, an accurate assessment of the relative position between the light source and the target is often unavailable in real situations. The Hayakawa procedure is a computational method for estimating such information directly from the data images. It occasionally breaks down when some of the available images excessively deviate from ideality. This is generally due to observing a non Lambertian surface, or illuminating it from a close distance, or both. Indeed, in narrow shooting scenarios, typical, e.g., of archaeological excavation sites, it is impossible to position a flashlight at a sufficient distance from the observed surface. It is then necessary to understand if a given dataset is reliable and which images should be selected to better reconstruct the target. In this paper, we propose some algorithms to perform this task and explore their effectiveness.

This article proposes a so-called Quasi Time-Reversible (QTR G-Ext) scheme based on Grassmann extrapolation of density matrices for an accurate calculation of initial guesses in Born-Oppenheimer Molecular Dynamics simulations. The method shows excellent results on four large molecular systems, ranging from 21 to 94 atoms simulated with Kohn-Sham density functional theory surrounded with a classical environment with 6k to 16k atoms. Namely, it clearly reduces the number of self-consistent field iterations, while keeping a similar energy drift as in the extended Lagrangian Born-Oppenheimer method.

The main technological obstacle hampering the dissemination of modern optoelectronic devices operating with large light-matter coupling strength ${\Omega}$ is an in-depth comprehension of the carrier current extraction and injection from and into strongly coupled light-matter states, the so-called polaritonic states. The main challenge lies in modeling the interaction between excitations of different nature, namely bosonic excitations (the plasmonic ISB excitations) with fermionic excitations (the electrons within the extraction or injection subband). In this work, we introduce a comprehensive quantum framework that encompasses both the ISB plasmonic mode and the extractor/injector mode, with a specific emphasis on accurately describing the coherent nature of transport. This reveals inherent selection rules dictating the interaction between the ISB plasmon and the extraction/injection subband. To incorporate the dynamics of the system, this framework is combined to a density matrix model and a quantum master equation which have the key property to distinguish intra and intersubband mechanisms. These theoretical developments are confronted to experimental photocurrent measurements from midinfrared quantum cascade detectors (${\lambda}$ = 10 ${\mu}$m) embedded in metal-semiconductor-metal microcavities, operating at the onset of the strong light-matter coupling regime (2${\Omega}$ = 9.3 meV). We are able to reproduce quantitatively the different features of the photocurrent spectra, notably the relative amplitude evolution of the polaritonic peaks with respect to the voltage bias applied to the structure. These results on extraction allow us to elucidate the possibility to effectively inject electronic excitations into ISB plasmonic states, and thus polaritonic states.

Recurrent neural networks trained with the backpropagation through time (BPTT) algorithm have led to astounding successes in various temporal tasks. However, BPTT introduces severe limitations, such as the requirement to propagate information backwards through time, the weight symmetry requirement, as well as update-locking in space and time. These problems become roadblocks for AI systems where online training capabilities are vital. Recently, researchers have developed biologically-inspired training algorithms, addressing a subset of those problems. In this work, we propose a novel learning algorithm called online spatio-temporal learning with target projection (OSTTP) that resolves all aforementioned issues of BPTT. In particular, OSTTP equips a network with the capability to simultaneously process and learn from new incoming data, alleviating the weight symmetry and update-locking problems. We evaluate OSTTP on two temporal tasks, showcasing competitive performance compared to BPTT. Moreover, we present a proof-of-concept implementation of OSTTP on a memristive neuromorphic hardware system, demonstrating its versatility and applicability to resource-constrained AI devices., Comment: Accepted at IEEE AICAS 2023. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media. Copyright 2023 IEEE

Electromagnetic induction (EMI) techniques are widely used in geophysical surveying. Their success is mainly due to their easy and fast data acquisition, but the effectiveness of data inversion is strongly influenced by the quality of sensed data, resulting from suiting the device configuration to the physical features of the survey site. Forward modelling is an essential tool to optimize this aspect and design a successful surveying campaign. In this paper, a new software tool for forward EMI modelling is introduced. It extends and complements an existing open-source package for EMI data inversion, and includes an interactive graphical user interface. Its use is explained by a theoretical introduction and demonstrated through a simulated case study. The nonlinear data inversion issue is briefly discussed and the inversion module of the package is extended by a new regularized minimal-norm algorithm.

Overdetermined systems of first kind integral equations appear in many applications. When the right-hand side is discretized, the resulting finite-data problem is ill-posed and admits infinitely many solutions. We propose a numerical method to compute the minimal-norm solution in the presence of boundary constraints. The algorithm stems from the Riesz representation theorem and operates in a reproducing kernel Hilbert space. Since the resulting linear system is strongly ill-conditioned, we construct a regularization method depending on a discrete parameter. It is based on the expansion of the minimal-norm solution in terms of the singular functions of the integral operator defining the problem. Two estimation techniques are tested for the automatic determination of the regularization parameter, namely, the discrepancy principle and the L-curve method. Numerical results concerning two artificial test problems demonstrate the excellent performance of the proposed method. Finally, a particular model typical of geophysical applications, which reproduces the readings of a frequency domain electromagnetic induction device, is investigated. The results show that the new method is extremely effective when the sought solution is smooth, but gives significant information on the solution even for non-smooth solutions.

We report results from a study of hadronic transitions of the $\chi_{bJ}(nP)$ states of bottomonium at Belle. The $P$-wave states are reconstructed in transitions to the $\Upsilon(1S)$ with the emission of an $\omega$ meson. The transitions of the $n=2$ triplet states provide a unique laboratory in which to study nonrelativistic quantum chromodynamics, as the kinematic threshold for production of an $\omega$ and $\Upsilon(1S)$ lies between the $J=0$ and $J=1$ states. A search for the $\chi_{bJ}(3P)$ states is also reported.

When a physical system is modeled by a nonlinear function, the unknown parameters can be estimated by fitting experimental observations by a least-squares approach. Newton's method and its variants are often used to solve problems of this type. In this paper, we are concerned with the computation of the minimal-norm solution of an underdetermined nonlinear least-squares problem. We present a Gauss-Newton type method, which relies on two relaxation parameters to ensure convergence, and which incorporates a procedure to dynamically estimate the two parameters, as well as the rank of the Jacobian matrix, along the iterations. Numerical results are presented.

Soon after its introduction in 2009, Bitcoin has been adopted by cyber-criminals, which rely on its pseudonymity to implement virtually untraceable scams. One of the typical scams that operate on Bitcoin are the so-called Ponzi schemes. These are fraudulent investments which repay users with the funds invested by new users that join the scheme, and implode when it is no longer possible to find new investments. Despite being illegal in many countries, Ponzi schemes are now proliferating on Bitcoin, and they keep alluring new victims, who are plundered of millions of dollars. We apply data mining techniques to detect Bitcoin addresses related to Ponzi schemes. Our starting point is a dataset of features of real-world Ponzi schemes, that we construct by analysing, on the Bitcoin blockchain, the transactions used to perform the scams. We use this dataset to experiment with various machine learning algorithms, and we assess their effectiveness through standard validation protocols and performance metrics. The best of the classifiers we have experimented can identify most of the Ponzi schemes in the dataset, with a low number of false positives.