Search

We consider the Higgs branch of 5d fixed points engineered using brane webs with an O7$^+$-plane. We use the brane construction to propose a set of rules to extract the corresponding magnetic quivers. Such magnetic quivers are generically framed non-simply-laced quivers containing unitary as well as special unitary gauge nodes. We compute the Coulomb branch Hilbert series of the proposed magnetic quivers. In some specific cases, an alternative magnetic quiver can be obtained either using an ordinary brane web or a brane web with an O5-plane. In these cases, we find a match at the level of the Hilbert series., Comment: 21 pages

This paper discusses the perspective of the H2020 TEACHING project on the next generation of autonomous applications running in a distributed and highly heterogeneous environment comprising both virtual and physical resources spanning the edge-cloud continuum. TEACHING puts forward a human-centred vision leveraging the physiological, emotional, and cognitive state of the users as a driver for the adaptation and optimization of the autonomous applications. It does so by building a distributed, embedded and federated learning system complemented by methods and tools to enforce its dependability, security and privacy preservation. The paper discusses the main concepts of the TEACHING approach and singles out the main AI-related research challenges associated with it. Further, we provide a discussion of the design choices for the TEACHING system to tackle the aforementioned challenges

Argyres-Douglas theories constitute an important class of superconformal field theories in $4$d. The main focus of this paper is on two infinite families of such theories, known as $D^b_p(\mathrm{SO}(2N))$ and $(A_m, D_n)$. We analyze in depth their conformal manifolds. In doing so we encounter several theories of class $\mathcal{S}$ of twisted $A_{\text{odd}}$, twisted $A_{\text{even}}$ and twisted $D$ types associated with a sphere with one twisted irregular puncture and one twisted regular puncture. These models include $D_p(G)$ theories, with $G$ non-simply-laced algebras. A number of new properties of such theories are discussed in detail, along with new SCFTs that arise from partially closing the twisted regular puncture. Moreover, we systematically present the $3$d mirror theories, also known as the magnetic quivers, for the $D^b_p(\mathrm{SO}(2N))$ theories, with $p \geq b$, and the $(A_m, D_n)$ theories, with arbitrary $m$ and $n$. We also discuss the $3$d reduction and mirror theories of certain $D^b_p(\mathrm{SO}(2N))$ theories, with $p < b$, where the former arises from gauging topological symmetries of some $T^\sigma_\rho[\mathrm{SO}(2M)]$ theories that are not manifest in the Lagrangian description of the latter., Comment: v2: added references. JHEP accepted. v1: 61 pages + 3 appendices

Continual Learning (CL) refers to a learning setup where data is non stationary and the model has to learn without forgetting existing knowledge. The study of CL for sequential patterns revolves around trained recurrent networks. In this work, instead, we introduce CL in the context of Echo State Networks (ESNs), where the recurrent component is kept fixed. We provide the first evaluation of catastrophic forgetting in ESNs and we highlight the benefits in using CL strategies which are not applicable to trained recurrent models. Our results confirm the ESN as a promising model for CL and open to its use in streaming scenarios., Comment: Accepted as oral at ESANN 2021

We introduce the IBM Analog Hardware Acceleration Kit, a new and first of a kind open source toolkit to simulate analog crossbar arrays in a convenient fashion from within PyTorch (freely available at https://github.com/IBM/aihwkit). The toolkit is under active development and is centered around the concept of an "analog tile" which captures the computations performed on a crossbar array. Analog tiles are building blocks that can be used to extend existing network modules with analog components and compose arbitrary artificial neural networks (ANNs) using the flexibility of the PyTorch framework. Analog tiles can be conveniently configured to emulate a plethora of different analog hardware characteristics and their non-idealities, such as device-to-device and cycle-to-cycle variations, resistive device response curves, and weight and output noise. Additionally, the toolkit makes it possible to design custom unit cell configurations and to use advanced analog optimization algorithms such as Tiki-Taka. Moreover, the backward and update behavior can be set to "ideal" to enable hardware-aware training features for chips that target inference acceleration only. To evaluate the inference accuracy of such chips over time, we provide statistical programming noise and drift models calibrated on phase-change memory hardware. Our new toolkit is fully GPU accelerated and can be used to conveniently estimate the impact of material properties and non-idealities of future analog technology on the accuracy for arbitrary ANNs., Comment: Submitted to AICAS2021

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 prototyping, training, and reproducible evaluation of continual learning algorithms., Comment: Official Website: https://avalanche.continualai.org

Replay strategies are Continual Learning techniques which mitigate catastrophic forgetting by keeping a buffer of patterns from previous experiences, which are interleaved with new data during training. The amount of patterns stored in the buffer is a critical parameter which largely influences the final performance and the memory footprint of the approach. This work introduces Distilled Replay, a novel replay strategy for Continual Learning which is able to mitigate forgetting by keeping a very small buffer (1 pattern per class) of highly informative samples. Distilled Replay builds the buffer through a distillation process which compresses a large dataset into a tiny set of informative examples. We show the effectiveness of our Distilled Replay against popular replay-based strategies on four Continual Learning benchmarks.

In this work, we study the phenomenon of catastrophic forgetting in the graph representation learning scenario. The primary objective of the analysis is to understand whether classical continual learning techniques for flat and sequential data have a tangible impact on performances when applied to graph data. To do so, we experiment with a structure-agnostic model and a deep graph network in a robust and controlled environment on three different datasets. The benchmark is complemented by an investigation on the effect of structure-preserving regularization techniques on catastrophic forgetting. We find that replay is the most effective strategy in so far, which also benefits the most from the use of regularization. Our findings suggest interesting future research at the intersection of the continual and graph representation learning fields. Finally, we provide researchers with a flexible software framework to reproduce our results and carry out further experiments., Comment: Accepted at the 2021 Web Conference Workshop on Graph Learning Benchmarks (GLB 2021). Code available at https://github.com/diningphil/continual_learning_for_graphs

Learning continuously during all model lifetime is fundamental to deploy machine learning solutions robust to drifts in the data distribution. Advances in Continual Learning (CL) with recurrent neural networks could pave the way to a large number of applications where incoming data is non stationary, like natural language processing and robotics. However, the existing body of work on the topic is still fragmented, with approaches which are application-specific and whose assessment is based on heterogeneous learning protocols and datasets. In this paper, we organize the literature on CL for sequential data processing by providing a categorization of the contributions and a review of the benchmarks. We propose two new benchmarks for CL with sequential data based on existing datasets, whose characteristics resemble real-world applications. We also provide a broad empirical evaluation of CL and Recurrent Neural Networks in class-incremental scenario, by testing their ability to mitigate forgetting with a number of different strategies which are not specific to sequential data processing. Our results highlight the key role played by the sequence length and the importance of a clear specification of the CL scenario., Comment: Published in Neural Networks

In the last years, scientific and industrial research has experienced a growing interest in acquiring large annotated data sets to train artificial intelligence algorithms for tackling problems in different domains. In this context, we have observed that even the market for football data has substantially grown. The analysis of football matches relies on the annotation of both individual players' and team actions, as well as the athletic performance of players. Consequently, annotating football events at a fine-grained level is a very expensive and error-prone task. Most existing semi-automatic tools for football match annotation rely on cameras and computer vision. However, those tools fall short in capturing team dynamics, and in extracting data of players who are not visible in the camera frame. To address these issues, in this manuscript we present FootApp, an AI-based system for football match annotation. First, our system relies on an advanced and mixed user interface that exploits both vocal and touch interaction. Second, the motor performance of players is captured and processed by applying machine learning algorithms to data collected from inertial sensors worn by players. Artificial intelligence techniques are then used to check the consistency of generated labels, including those regarding the physical activity of players, to automatically recognize annotation errors. Notably, we implemented a full prototype of the proposed system, performing experiments to show its effectiveness in a real-world adoption scenario.

Since their appearance, Smart Cities have aimed at improving the daily life of people, helping to make public services smarter and more efficient. Several of these services are often intended to provide better security conditions for citizens and drivers. In this vein, we present Heimdall, an AI-based video surveillance system for traffic monitoring and anomalies detection. The proposed system features three main tiers: a ground level, consisting of a set of smart lampposts equipped with cameras and sensors, and an advanced AI unit for detecting accidents and traffic anomalies in real time; a territorial level, which integrates and combines the information collected from the different lampposts, and cross-correlates it with external data sources, in order to coordinate and handle warnings and alerts; a training level, in charge of continuously improving the accuracy of the modules that have to sense the environment. Finally, we propose and discuss an early experimental approach for the detection of anomalies, based on a Faster R-CNN, and adopted in the proposed infrastructure.

We study the moduli space of 3d $\mathcal{N}=4$ quiver gauge theories with unitary, orthogonal and symplectic gauge nodes, that fall into exceptional sequences. We find that both the Higgs and Coulomb branches of the moduli space factorise into decoupled sectors. Each decoupled sector is described by a single quiver gauge theory with only unitary gauge nodes. The orthosymplectic quivers serve as magnetic quivers for 5d $\mathcal{N}=1$ superconformal field theories which can be engineered in type IIB string theories both with and without an O5 plane. We use this point of view to postulate the dual pairs of unitary and orthosymplectic quivers by deriving them as magnetic quivers of the 5d theory. We use this correspondence to conjecture exact highest weight generating functions for the Coulomb branch Hilbert series of the orthosymplectic quivers, and provide tests of these results by directly computing the Hilbert series for the orthosymplectic quivers in a series expansion., Comment: 46 pages, 10 tables, many pictures, v2: references added, typos corrected, v3: Appendix E added, matches published version

We propose a combined mechanism to realize both winding inflation and de Sitter uplifts. We realize the necessary structure of competing terms in the scalar potential not via tuning the vacuum expectation values of the complex structure moduli, but by a hierarchy of the Gopakumar-Vafa invariants of the underlying Calabi-Yau threefold. To show that Calabi-Yau threefolds with the prescribed hierarchy actually exist, we explicitly create a database of all the genus $0$ Gopakumar-Vafa invariants up to total degree $10$ for all the complete intersection Calabi-Yau's up to Picard number $9$. As a side product, we also identify all the redundancies present in the CICY list, up to Picard number $13$. Both databases can be accessed at this link: https://www.desy.de/~westphal/GV_CICY_webpage/GVInvariants.html ., Comment: v2, JHEP accepted, added references, 1+44 pages, 5 figures, 3 appendices

In flux compactifications of type IIB string theory with D3 and seven-branes, the negative induced D3 charge localized on seven-branes leads to an apparently pathological profile of the metric sufficiently close to the source. With the volume modulus stabilized in a KKLT de Sitter vacuum this pathological region takes over a significant part of the entire compactification, threatening to spoil the KKLT effective field theory. In this paper we employ the Seiberg-Witten solution of pure $SU(N)$ super Yang-Mills theory to argue that wrapped seven-branes can be thought of as bound states of more microscopic exotic branes. We argue that the low-energy worldvolume dynamics of a stack of $n$ such exotic branes is given by the $(A_1,A_{n-1})$ Argyres-Douglas theory. Moreover, the splitting of the perturbative (in $\alpha'$) seven-brane into its constituent branes at the non-perturbative level resolves the apparently pathological region close to the seven-brane and replaces it with a region of $\mathcal{O}(1)$ Einstein frame volume. While this region generically takes up an $\mathcal{O}(1)$ fraction of the compactification in a KKLT de Sitter vacuum we argue that a small flux superpotential \textit{dynamically} ensures that the 4d effective field theory of KKLT remains valid nevertheless., Comment: 42 pages, 19 figures, 1 appendix; v2: references added, minor corrections; v3: monodromies corrected, reference added; v4: matches version published in JHEP

Training RNNs to learn long-term dependencies is difficult due to vanishing gradients. We explore an alternative solution based on explicit memorization using linear autoencoders for sequences, which allows to maximize the short-term memory and that can be solved with a closed-form solution without backpropagation. We introduce an initialization schema that pretrains the weights of a recurrent neural network to approximate the linear autoencoder of the input sequences and we show how such pretraining can better support solving hard classification tasks with long sequences. We test our approach on sequential and permuted MNIST. We show that the proposed approach achieves a much lower reconstruction error for long sequences and a better gradient propagation during the finetuning phase., Comment: Accepted at NeurIPS 2020 workshop "Beyond Backpropagation: Novel Ideas for Training Neural Architectures"

We present VisualHints, a novel environment for multimodal reinforcement learning (RL) involving text-based interactions along with visual hints (obtained from the environment). Real-life problems often demand that agents interact with the environment using both natural language information and visual perception towards solving a goal. However, most traditional RL environments either solve pure vision-based tasks like Atari games or video-based robotic manipulation; or entirely use natural language as a mode of interaction, like Text-based games and dialog systems. In this work, we aim to bridge this gap and unify these two approaches in a single environment for multimodal RL. We introduce an extension of the TextWorld cooking environment with the addition of visual clues interspersed throughout the environment. The goal is to force an RL agent to use both text and visual features to predict natural language action commands for solving the final task of cooking a meal. We enable variations and difficulties in our environment to emulate various interactive real-world scenarios. We present a baseline multimodal agent for solving such problems using CNN-based feature extraction from visual hints and LSTMs for textual feature extraction. We believe that our proposed visual-lingual environment will facilitate novel problem settings for the RL community., Comment: Code is available at http://ibm.biz/VisualHints

The polymer of intrinsic microporosity PIM 1 is employed to disperse and deposit a Pt at titanate nanosheet photocatalyst film. The resulting microporous films allow electrolyte and oxygen permeation to give conventional oxygen reduction voltammetric responses on glassy carbon or on platinum disk electrodes in the dark. Preliminary data are presented showing that with pulsed light from a blue LED oxygen reduction at the electrode is effectively switched off. A mechanism is tentatively assigned as photocatalytic depletion of oxygen near the electrode. Photoelectrochemical current responses are observed in aqueous NaOH, NaCl, Na2HPO4 and shown to be light intensity and oxygen concentration dependent. Electroanalytical applications are suggested.

We study the Higgs branch of 5d superconformal theories engineered from brane webs with orientifold five-planes. We propose a generalization of the rules to derive magnetic quivers from brane webs pioneered in arXiv:2004.04082, by analyzing theories that can be described with a brane web with and without O5 planes. Our proposed magnetic quivers include novel features, such as hypermultiplets transforming in the fundamental-fundamental representation of two gauge nodes, antisymmetric matter, and $\mathbb{Z}_2$ gauge nodes. We test our results by computing the Coulomb and Higgs branch Hilbert series of the magnetic quivers obtained from the two distinct constructions and find agreement in all cases., Comment: 67 pages, 57 figures, 24 tables, 4 appendices, minor corrections, matches with the published version

We present a mechanism for realizing hybrid inflation using two axion fields with a purely non-perturbatively generated scalar potential. The structure of scalar potential is highly constrained by the discrete shift symmetries of the axions. We show that harmonic hybrid inflation generates observationally viable slow-roll inflation for a wide range of initial conditions. This is possible while accommodating certain UV arguments favoring constraints $f\lesssim M_{\rm P}$ and $\Delta\phi_{60}\lesssim M_{\rm P}$ on the axion periodicity and slow-roll field range, respectively. We discuss controlled $\mathbb{Z}_2$-symmetry breaking of the adjacent axion vacua as a means of avoiding cosmological domain wall problems. Including a minimal form of $\mathbb{Z}_2$-symmetry breaking into the minimally tuned setup leads to a prediction of primordial tensor modes with the tensor-to-scalar ratio in the range $10^{-4}\lesssim r \lesssim 0.01$, directly accessible to upcoming CMB observations. Finally, we outline several avenues towards realizing harmonic hybrid inflation in type IIB string theory., Comment: 16 pages, 7 figures, 1 appendix, LaTeX. V2 accepted in JHEP

The effectiveness of recurrent neural networks can be largely influenced by their ability to store into their dynamical memory information extracted from input sequences at different frequencies and timescales. Such a feature can be introduced into a neural architecture by an appropriate modularization of the dynamic memory. In this paper we propose a novel incrementally trained recurrent architecture targeting explicitly multi-scale learning. First, we show how to extend the architecture of a simple RNN by separating its hidden state into different modules, each subsampling the network hidden activations at different frequencies. Then, we discuss a training algorithm where new modules are iteratively added to the model to learn progressively longer dependencies. Each new module works at a slower frequency than the previous ones and it is initialized to encode the subsampled sequence of hidden activations. Experimental results on synthetic and real-world datasets on speech recognition and handwritten characters show that the modular architecture and the incremental training algorithm improve the ability of recurrent neural networks to capture long-term dependencies., Comment: accepted @ ECML 2020. arXiv admin note: substantial text overlap with arXiv:2001.11771

The ability to learn in dynamic, nonstationary environments without forgetting previous knowledge, also known as Continual Learning (CL), is a key enabler for scalable and trustworthy deployments of adaptive solutions. While the importance of continual learning is largely acknowledged in machine vision and reinforcement learning problems, this is mostly under-documented for sequence processing tasks. This work proposes a Recurrent Neural Network (RNN) model for CL that is able to deal with concept drift in input distribution without forgetting previously acquired knowledge. We also implement and test a popular CL approach, Elastic Weight Consolidation (EWC), on top of two different types of RNNs. Finally, we compare the performances of our enhanced architecture against EWC and RNNs on a set of standard CL benchmarks, adapted to the sequential data processing scenario. Results show the superior performance of our architecture and highlight the need for special solutions designed to address CL in RNNs., Comment: Accepted as a conference paper at 2020 International Joint Conference on Neural Networks (IJCNN 2020). Part of 2020 IEEE World Congress on Computational Intelligence (IEEE WCCI 2020)

We present a vast landscape of O3/O7 orientifolds that descends from the famous set of complete intersection Calabi-Yau threefolds (CICY). We give distributions of topological data relevant for phenomenology such as the orientifold-odd Hodge numbers, the D3-tadpole, and multiplicities of O3 and O7-planes. Somewhat surprisingly, almost all of these orientifolds have conifold singularities whose deformation branches are projected out by the orientifolding. However, they can be resolved, so most of the orientifolds actually descend from a much larger and possibly new set of CY threefolds that can be reached from the CICYs via conifold transitions. We observe an interesting class of $\mathcal{N}=1$ geometric transitions involving colliding O-planes. Finally, as an application, we use our dataset to produce examples of orientifolds that satisfy the topological requirements for the existence of ultra-light throat axions (\textit{thraxions}) as proposed in \cite{Hebecker:2018yxs}. The database can be accessed at https://www.desy.de/~westphal/orientifold_webpage/cicy_orientifolds.html, Comment: 35 pages, 3 appendices, 5 figures

We prove that a very large class of $15502$ general Argyres-Douglas theories cannot admit a UV lagrangian which flows to them via the Maruyoshi-Song supersymmetry enhancement mechanism. We do so by developing a computer program which brute-force lists, for any given 4d $\mathcal{N}=2$ superconformal theory $\mathcal{T}_{\text{IR}}$, all possible UV candidate superconformal lagrangians $\mathcal{T}_{\text{UV}}$ satisfying some necessary criteria for the supersymmetry enhancement to happen. We argue that this is enough evidence to conjecture that it is impossible, in general, to find new examples of Maruyoshi-Song lagrangians for generalized Argyres-Douglas theories. All lagrangians already known are, on the other hand, recovered and confirmed in our scan. Finally, we also develop another program to compute efficiently Coulomb branch spectrum, masses, couplings and central charges for $(G,G')$ Argyres-Douglas theories of arbitrarily high rank., Comment: 37 pages, 2 figures, 15 tables, 3 ancillary files. version 2, accepted in JHEP