-
Automated Behavioral Analysis Using Instance Segmentation
Authors:
Chen Yang,
Jeremy Forest,
Matthew Einhorn,
Thomas A. Cleland
Abstract:
Animal behavior analysis plays a crucial role in various fields, such as life science and biomedical research. However, the scarcity of available data and the high cost associated with obtaining a large number of labeled datasets pose significant challenges. In this research, we propose a novel approach that leverages instance segmentation-based transfer learning to address these issues. By capita…
▽ More
Animal behavior analysis plays a crucial role in various fields, such as life science and biomedical research. However, the scarcity of available data and the high cost associated with obtaining a large number of labeled datasets pose significant challenges. In this research, we propose a novel approach that leverages instance segmentation-based transfer learning to address these issues. By capitalizing on fine-tuning the classification head of the instance segmentation network, we enable the tracking of multiple animals and facilitate behavior analysis in laboratory-recorded videos. To demonstrate the effectiveness of our method, we conducted a series of experiments, revealing that our approach achieves exceptional performance levels, comparable to human capabilities, across a diverse range of animal behavior analysis tasks. Moreover, we emphasize the practicality of our solution, as it requires only a small number of labeled images for training. To facilitate the adoption and further development of our method, we have developed an open-source implementation named Annolid (An annotation and instance segmentation-based multiple animal tracking and behavior analysis package). The codebase is publicly available on GitHub at https://github.com/cplab/annolid. This resource serves as a valuable asset for researchers and practitioners interested in advancing animal behavior analysis through state-of-the-art techniques.
△ Less
Submitted 12 December, 2023;
originally announced December 2023.
-
NeuroBench: A Framework for Benchmarking Neuromorphic Computing Algorithms and Systems
Authors:
Jason Yik,
Korneel Van den Berghe,
Douwe den Blanken,
Younes Bouhadjar,
Maxime Fabre,
Paul Hueber,
Denis Kleyko,
Noah Pacik-Nelson,
Pao-Sheng Vincent Sun,
Guangzhi Tang,
Shenqi Wang,
Biyan Zhou,
Soikat Hasan Ahmed,
George Vathakkattil Joseph,
Benedetto Leto,
Aurora Micheli,
Anurag Kumar Mishra,
Gregor Lenz,
Tao Sun,
Zergham Ahmed,
Mahmoud Akl,
Brian Anderson,
Andreas G. Andreou,
Chiara Bartolozzi,
Arindam Basu
, et al. (73 additional authors not shown)
Abstract:
Neuromorphic computing shows promise for advancing computing efficiency and capabilities of AI applications using brain-inspired principles. However, the neuromorphic research field currently lacks standardized benchmarks, making it difficult to accurately measure technological advancements, compare performance with conventional methods, and identify promising future research directions. Prior neu…
▽ More
Neuromorphic computing shows promise for advancing computing efficiency and capabilities of AI applications using brain-inspired principles. However, the neuromorphic research field currently lacks standardized benchmarks, making it difficult to accurately measure technological advancements, compare performance with conventional methods, and identify promising future research directions. Prior neuromorphic computing benchmark efforts have not seen widespread adoption due to a lack of inclusive, actionable, and iterative benchmark design and guidelines. To address these shortcomings, we present NeuroBench: a benchmark framework for neuromorphic computing algorithms and systems. NeuroBench is a collaboratively-designed effort from an open community of nearly 100 co-authors across over 50 institutions in industry and academia, aiming to provide a representative structure for standardizing the evaluation of neuromorphic approaches. The NeuroBench framework introduces a common set of tools and systematic methodology for inclusive benchmark measurement, delivering an objective reference framework for quantifying neuromorphic approaches in both hardware-independent (algorithm track) and hardware-dependent (system track) settings. In this article, we present initial performance baselines across various model architectures on the algorithm track and outline the system track benchmark tasks and guidelines. NeuroBench is intended to continually expand its benchmarks and features to foster and track the progress made by the research community.
△ Less
Submitted 17 January, 2024; v1 submitted 10 April, 2023;
originally announced April 2023.
-
Space Plasma Physics Science Opportunities for the Lunar Orbital Platform -Gateway
Authors:
Iannis Dandouras,
Matt G G T Taylor,
Johan de Keyser,
Yoshifumi Futaana,
Ruth A Bamford,
Graziella Branduardi-Raymont,
Jean-Yves Chaufray,
Dragos Constantinescu,
Elisabetta de Angelis,
Pierre Devoto,
Jonathan Eastwood,
Marius Echim,
Philippe Garnier,
Benjamin Grison,
David Hercik,
Helmut Lammer,
André Laurens,
François Leblanc,
Anna Milillo,
Rumi Nakamura,
Lubomír Přech,
Elias Roussos,
Štěpán Štverák,
Julien Forest,
Arnaud Trouche
, et al. (4 additional authors not shown)
Abstract:
The Lunar Orbital Platform-Gateway (LOP-Gateway, or simply Gateway) is a crewed platform that will be assembled and operated in the vicinity of the Moon by NASA and international partner organizations, including ESA, starting from the mid-2020s. It will offer new opportunities for fundamental and applied scientific research. The Moon is a unique location to study the deep space plasma environment.…
▽ More
The Lunar Orbital Platform-Gateway (LOP-Gateway, or simply Gateway) is a crewed platform that will be assembled and operated in the vicinity of the Moon by NASA and international partner organizations, including ESA, starting from the mid-2020s. It will offer new opportunities for fundamental and applied scientific research. The Moon is a unique location to study the deep space plasma environment. Moreover, the lunar surface and the surface-bounded exosphere are interacting with this environment, constituting a complex multi-scale interacting system. This paper examines the opportunities provided by externally mounted payloads on the Gateway in the field of space plasma physics, heliophysics and space weather, but also examines the impact of the space environment on an inhabited platform in the vicinity of the Moon. It then presents the conceptual design of a model payload, required to perform these space plasma measurements and observations. It results that the Gateway is very well-suited for space plasma physics research. It allows a series of scientific objectives with a multidisciplinary dimension to be addressed.
△ Less
Submitted 20 December, 2022;
originally announced January 2023.
-
Challenging the Universal Representation of Deep Models for 3D Point Cloud Registration
Authors:
David Bojanić,
Kristijan Bartol,
Josep Forest,
Stefan Gumhold,
Tomislav Petković,
Tomislav Pribanić
Abstract:
Learning universal representations across different applications domain is an open research problem. In fact, finding universal architecture within the same application but across different types of datasets is still unsolved problem too, especially in applications involving processing 3D point clouds. In this work we experimentally test several state-of-the-art learning-based methods for 3D point…
▽ More
Learning universal representations across different applications domain is an open research problem. In fact, finding universal architecture within the same application but across different types of datasets is still unsolved problem too, especially in applications involving processing 3D point clouds. In this work we experimentally test several state-of-the-art learning-based methods for 3D point cloud registration against the proposed non-learning baseline registration method. The proposed method either outperforms or achieves comparable results w.r.t. learning based methods. In addition, we propose a dataset on which learning based methods have a hard time to generalize. Our proposed method and dataset, along with the provided experiments, can be used in further research in studying effective solutions for universal representations. Our source code is available at: github.com/DavidBoja/greedy-grid-search.
△ Less
Submitted 29 November, 2022;
originally announced November 2022.
-
Avoiding Catastrophe: Active Dendrites Enable Multi-Task Learning in Dynamic Environments
Authors:
Abhiram Iyer,
Karan Grewal,
Akash Velu,
Lucas Oliveira Souza,
Jeremy Forest,
Subutai Ahmad
Abstract:
A key challenge for AI is to build embodied systems that operate in dynamically changing environments. Such systems must adapt to changing task contexts and learn continuously. Although standard deep learning systems achieve state of the art results on static benchmarks, they often struggle in dynamic scenarios. In these settings, error signals from multiple contexts can interfere with one another…
▽ More
A key challenge for AI is to build embodied systems that operate in dynamically changing environments. Such systems must adapt to changing task contexts and learn continuously. Although standard deep learning systems achieve state of the art results on static benchmarks, they often struggle in dynamic scenarios. In these settings, error signals from multiple contexts can interfere with one another, ultimately leading to a phenomenon known as catastrophic forgetting. In this article we investigate biologically inspired architectures as solutions to these problems. Specifically, we show that the biophysical properties of dendrites and local inhibitory systems enable networks to dynamically restrict and route information in a context-specific manner. Our key contributions are as follows. First, we propose a novel artificial neural network architecture that incorporates active dendrites and sparse representations into the standard deep learning framework. Next, we study the performance of this architecture on two separate benchmarks requiring task-based adaptation: Meta-World, a multi-task reinforcement learning environment where a robotic agent must learn to solve a variety of manipulation tasks simultaneously; and a continual learning benchmark in which the model's prediction task changes throughout training. Analysis on both benchmarks demonstrates the emergence of overlap** but distinct and sparse subnetworks, allowing the system to fluidly learn multiple tasks with minimal forgetting. Our neural implementation marks the first time a single architecture has achieved competitive results on both multi-task and continual learning settings. Our research sheds light on how biological properties of neurons can inform deep learning systems to address dynamic scenarios that are typically impossible for traditional ANNs to solve.
△ Less
Submitted 25 April, 2022; v1 submitted 31 December, 2021;
originally announced January 2022.
-
Avalanche: an End-to-End Library for Continual Learning
Authors:
Vincenzo Lomonaco,
Lorenzo Pellegrini,
Andrea Cossu,
Antonio Carta,
Gabriele Graffieti,
Tyler L. Hayes,
Matthias De Lange,
Marc Masana,
Jary Pomponi,
Gido van de Ven,
Martin Mundt,
Qi She,
Keiland Cooper,
Jeremy Forest,
Eden Belouadah,
Simone Calderara,
German I. Parisi,
Fabio Cuzzolin,
Andreas Tolias,
Simone Scardapane,
Luca Antiga,
Subutai Amhad,
Adrian Popescu,
Christopher Kanan,
Joost van de Weijer
, et al. (3 additional authors not shown)
Abstract:
Learning continually from non-stationary data streams is a long-standing goal and a challenging problem in machine learning. Recently, we have witnessed a renewed and fast-growing interest in continual learning, especially within the deep learning community. However, algorithmic solutions are often difficult to re-implement, evaluate and port across different settings, where even results on standa…
▽ More
Learning continually from non-stationary data streams is a long-standing goal and a challenging problem in machine learning. Recently, we have witnessed a renewed and fast-growing interest in continual learning, especially within the deep learning community. However, algorithmic solutions are often difficult to re-implement, evaluate and port across different settings, where even results on standard benchmarks are hard to reproduce. In this work, we propose Avalanche, an open-source end-to-end library for continual learning research based on PyTorch. Avalanche is designed to provide a shared and collaborative codebase for fast prototy**, training, and reproducible evaluation of continual learning algorithms.
△ Less
Submitted 1 April, 2021;
originally announced April 2021.
-
Effects of Nonlocal One-Pion-Exchange Potential in Deuteron
Authors:
J. L. Forest
Abstract:
The off-shell aspects of the one-pion-exchange potential (OPEP) are discussed. Relativistic Hamiltonians containing relativistic kinetic energy, relativistic OPEP with various off-shell behaviors and Argonne $v_{18}$ short-range parameterization are used to study the deuteron properties. The OPEP off-shell behaviors depend on whether a pseudovector or pseudoscalar pion-nucleon coupling is used a…
▽ More
The off-shell aspects of the one-pion-exchange potential (OPEP) are discussed. Relativistic Hamiltonians containing relativistic kinetic energy, relativistic OPEP with various off-shell behaviors and Argonne $v_{18}$ short-range parameterization are used to study the deuteron properties. The OPEP off-shell behaviors depend on whether a pseudovector or pseudoscalar pion-nucleon coupling is used and are characterized by a parameter $μ$. We study potentials having $μ$=-1, 0 and +1 and we find that they are nearly unitarily equivalent. We also find that a nonrelativistic Hamiltonian containing local potentials and nonrelativistic kinetic energy provides a good approximation to a Hamiltonian containing a relativistic OPEP based on pseudovector pion-nucleon coupling and relativistic kinetic energy.
△ Less
Submitted 28 May, 1999;
originally announced May 1999.
-
Quantum Monte Carlo Studies of Relativistic Effects in Light Nuclei
Authors:
J. L. Forest,
V. R. Pandharipande,
A. Arriaga
Abstract:
Relativistic Hamiltonians are defined as the sum of relativistic one-body kinetic energy, two- and three-body potentials and their boost corrections. In this work we use the variational Monte Carlo method to study two kinds of relativistic effects in the binding energy of 3H and 4He. The first is due to the nonlocalities in the relativistic kinetic energy and relativistic one-pion exchange poten…
▽ More
Relativistic Hamiltonians are defined as the sum of relativistic one-body kinetic energy, two- and three-body potentials and their boost corrections. In this work we use the variational Monte Carlo method to study two kinds of relativistic effects in the binding energy of 3H and 4He. The first is due to the nonlocalities in the relativistic kinetic energy and relativistic one-pion exchange potential (OPEP), and the second is from boost interaction. The OPEP contribution is reduced by about 15% by the relativistic nonlocality, which may also have significant effects on pion exchange currents. However, almost all of this reduction is canceled by changes in the kinetic energy and other interaction terms, and the total effect of the nonlocalities on the binding energy is very small. The boost interactions, on the other hand, give repulsive contributions of 0.4 (1.9) MeV in 3H (4He) and account for 37% of the phenomenological part of the three-nucleon interaction needed in the nonrelativistic Hamiltonians.
△ Less
Submitted 19 May, 1998;
originally announced May 1998.
-
Femtometer Toroidal Structures in Nuclei
Authors:
J. L. Forest,
V. R. Pandharipande,
Steven C. Pieper,
R. B. Wiringa,
R. Schiavilla,
A. Arriaga
Abstract:
The two-nucleon density distributions in states with isospin $T=0$, spin $S$=1 and projection $M_S$=0 and $\pm$1 are studied in $^2$H, $^{3,4}$He, $^{6,7}$Li and $^{16}$O. The equidensity surfaces for $M_S$=0 distributions are found to be toroidal in shape, while those of $M_S$=$\pm$1 have dumbbell shapes at large density. The dumbbell shapes are generated by rotating tori. The toroidal shapes i…
▽ More
The two-nucleon density distributions in states with isospin $T=0$, spin $S$=1 and projection $M_S$=0 and $\pm$1 are studied in $^2$H, $^{3,4}$He, $^{6,7}$Li and $^{16}$O. The equidensity surfaces for $M_S$=0 distributions are found to be toroidal in shape, while those of $M_S$=$\pm$1 have dumbbell shapes at large density. The dumbbell shapes are generated by rotating tori. The toroidal shapes indicate that the tensor correlations have near maximal strength at $r<2$ fm in all these nuclei. They provide new insights and simple explanations of the structure and electromagnetic form factors of the deuteron, the quasi-deuteron model, and the $dp$, $dd$ and $αd$ $L$=2 ($D$-wave) components in $^3$He, $^4$He and $^6$Li. The toroidal distribution has a maximum-density diameter of $\sim$1 fm and a half-maximum density thickness of $\sim$0.9 fm. Many realistic models of nuclear forces predict these values, which are supported by the observed electromagnetic form factors of the deuteron, and also predicted by classical Skyrme effective Lagrangians, related to QCD in the limit of infinite colors. Due to the rather small size of this structure, it could have a revealing relation to certain aspects of QCD.
△ Less
Submitted 21 March, 1996;
originally announced March 1996.
-
Variational Monte Carlo Calculations of $^3$H and $^4$He with a relativistic Hamiltonian - II
Authors:
J. L. Forest,
V. R. Pandharipande,
J. Carlson,
R. Schiavilla
Abstract:
In relativistic Hamiltonians the two-nucleon interaction is expressed as a sum of $\tilde{v}_{ij}$, the interaction in the ${\bf P}_{ij}=0$ rest frame, and the ``boost interaction'' $δv({\bf P}_{ij})$ which depends upon the total momentum ${\bf P}_{ij}$ and vanishes in the rest frame. The $δv$ can be regarded as a sum of four terms: $δv_{RE}$, $δv_{LC}$, $δv_{TP}$ and $δv_{QM}$; the first three…
▽ More
In relativistic Hamiltonians the two-nucleon interaction is expressed as a sum of $\tilde{v}_{ij}$, the interaction in the ${\bf P}_{ij}=0$ rest frame, and the ``boost interaction'' $δv({\bf P}_{ij})$ which depends upon the total momentum ${\bf P}_{ij}$ and vanishes in the rest frame. The $δv$ can be regarded as a sum of four terms: $δv_{RE}$, $δv_{LC}$, $δv_{TP}$ and $δv_{QM}$; the first three originate from the relativistic energy-momentum relation, Lorentz contraction and Thomas precession, while the last is purely quantum. The contributions of $δv_{RE}$ and $δv_{LC}$ have been previously calculated with the variational Monte Carlo method for $^3$H and $^4$He. In this brief note we report the results of similar calculations for the contributions of $δv_{TP}$ and $δv_{QM}$. These are found to be rather small.
△ Less
Submitted 5 October, 1994;
originally announced October 1994.
-
Pedagogical Studies of Relativistic Hamiltonians
Authors:
J. L. Forest,
V. R. Pandharipande,
J. L. Friar
Abstract:
Relativistic Hamiltonians are defined as the sum of relativistic one-body kinetic energies, two- and many-body interactions and their boost corrections. We review the calculation of the boost correction of the two-body interaction from commutation relations of the Poincare group and show that its important terms can be easily understood from classical relativistic mechanics. The boost correction…
▽ More
Relativistic Hamiltonians are defined as the sum of relativistic one-body kinetic energies, two- and many-body interactions and their boost corrections. We review the calculation of the boost correction of the two-body interaction from commutation relations of the Poincare group and show that its important terms can be easily understood from classical relativistic mechanics. The boost corrections for scalar- and vector- meson-exchange interactions, obtained from relativistic field theory, are shown to be in agreement with the results of the classical calculation. These boost corrections are also shown to be necessary to reproduce the known results of relativistic mean-field theories. We conclude with comments on the relativistic boost operator for the wave function of a nucleus. Several of the results presented in this pedagogical article are known. We hope that a better understanding of relativistic Hamiltonians and their relation to relativistic field theory is obtained by putting them together with a few new relations.
△ Less
Submitted 5 October, 1994;
originally announced October 1994.
