Search

Denoising Diffusion Probabilistic Models (DDPMs) have made great strides in generating high-quality samples in both discrete and continuous domains. However, Discrete DDPMs (D3PMs) have yet to be applied to the domain of Symbolic Music. This work presents the direct generation of Polyphonic Symbolic Music using D3PMs. Our model exhibits state-of-the-art sample quality, according to current quantitative evaluation metrics, and allows for flexible infilling at the note level. We further show, that our models are accessible to post-hoc classifier guidance, widening the scope of possible applications. However, we also cast a critical view on quantitative evaluation of music sample quality via statistical metrics, and present a simple algorithm that can confound our metrics with completely spurious, non-musical samples., Comment: In Proceedings of the 32nd International Joint Conference on Artificial Intelligence (IJCAI-23), Macau, China

The ability to generalize to a wide range of recording devices is a crucial performance factor for audio classification models. The characteristics of different types of microphones introduce distributional shifts in the digitized audio signals due to their varying frequency responses. If this domain shift is not taken into account during training, the model's performance could degrade severely when it is applied to signals recorded by unseen devices. In particular, training a model on audio signals recorded with a small number of different microphones can make generalization to unseen devices difficult. To tackle this problem, we convolve audio signals in the training set with pre-recorded device impulse responses (DIRs) to artificially increase the diversity of recording devices. We systematically study the effect of DIR augmentation on the task of Acoustic Scene Classification using CNNs and Audio Spectrogram Transformers. The results show that DIR augmentation in isolation performs similarly to the state-of-the-art method Freq-MixStyle. However, we also show that DIR augmentation and Freq-MixStyle are complementary, achieving a new state-of-the-art performance on signals recorded by devices unseen during training., Comment: In Proceedings of the 31st European Signal Processing Conference, EUSIPCO 2023. Source Code available at: https://github.com/theMoro/DIRAugmentation/

This paper targets the perceptual task of separating the different interacting voices, i.e., monophonic melodic streams, in a polyphonic musical piece. We target symbolic music, where notes are explicitly encoded, and model this task as a Multi-Trajectory Tracking (MTT) problem from discrete observations, i.e., notes in a pitch-time space. Our approach builds a graph from a musical piece, by creating one node for every note, and separates the melodic trajectories by predicting a link between two notes if they are consecutive in the same voice/stream. This kind of local, greedy prediction is made possible by node embeddings created by a heterogeneous graph neural network that can capture inter- and intra-trajectory information. Furthermore, we propose a new regularization loss that encourages the output to respect the MTT premise of at most one incoming and one outgoing link for every node, favouring monophonic (voice) trajectories; this loss function might also be useful in other general MTT scenarios. Our approach does not use domain-specific heuristics, is scalable to longer sequences and a higher number of voices, and can handle complex cases such as voice inversions and overlaps. We reach new state-of-the-art results for the voice separation task in classical music of different styles., Comment: Accepted at the 32nd International Joint Conference on Artificial Intelligence (IJCAI-23)

This paper introduces the ACCompanion, an expressive accompaniment system. Similarly to a musician who accompanies a soloist playing a given musical piece, our system can produce a human-like rendition of the accompaniment part that follows the soloist's choices in terms of tempo, dynamics, and articulation. The ACCompanion works in the symbolic domain, i.e., it needs a musical instrument capable of producing and playing MIDI data, with explicitly encoded onset, offset, and pitch for each played note. We describe the components that go into such a system, from real-time score following and prediction to expressive performance generation and online adaptation to the expressive choices of the human player. Based on our experience with repeated live demonstrations in front of various audiences, we offer an analysis of the challenges of combining these components into a system that is highly reactive and precise, while still a reliable musical partner, robust to possible performance errors and responsive to expressive variations., Comment: In Proceedings of the 32nd International Joint Conference on Artificial Intelligence (IJCAI-23), Macao, China. The differences/extensions with the previous version include a technical appendix, added missing links, and minor text updates. 10 pages, 4 figures

Wideband millimeter-wave communication systems can be extended to provide radar-like sensing capabilities on top of data communication, in a cost-effective manner. However, the development of joint communication and sensing technology is hindered by practical challenges, such as occlusions to the line-of-sight path and clock asynchrony between devices. The latter introduces time-varying timing and frequency offsets that prevent the estimation of sensing parameters and, in turn, the use of standard signal processing solutions. Existing approaches cannot be applied to commonly used phased-array receivers, as they build on stringent assumptions about the multipath environment, and are computationally complex. We present JUMP, the first system enabling practical bistatic and asynchronous joint communication and sensing, while achieving accurate target tracking and micro-Doppler extraction in realistic conditions. Our system compensates for the timing offset by exploiting the channel correlation across subsequent packets. Further, it tracks multipath reflections and eliminates frequency offsets by observing the phase of a dynamically-selected static reference path. JUMP has been implemented on a 60 GHz experimental platform, performing extensive evaluations of human motion sensing, including non-line-of-sight scenarios. In our results, JUMP attains comparable tracking performance to a full-duplex monostatic system and similar micro-Doppler quality with respect to a phase-locked bistatic receiver., Comment: 17 pages, 18 figures

This paper contributes tail bounds of the age-of-information of a general class of parallel systems and explores their potential. Parallel systems arise in relevant cases, such as in multi-band mobile networks, multi-technology wireless access, or multi-path protocols, just to name a few. Typically, control over each communication channel is limited and random service outages and congestion cause buffering that impairs the age-of-information. The parallel use of independent channels promises a remedy, since outages on one channel may be compensated for by another. Surprisingly, for the well-known case of M$\mid$M$\mid$1 queues we find the opposite: pooling capacity in one channel performs better than a parallel system with the same total capacity. A generalization is not possible since there are no solutions for other types of parallel queues at hand. In this work, we prove a dual representation of age-of-information in min-plus algebra that connects to queueing models known from the theory of effective bandwidth/capacity and the stochastic network calculus. Exploiting these methods, we derive tail bounds of the age-of-information of parallel G$\mid$G$\mid$1 queues. In addition to parallel classical queues, we investigate Markov channels where, depending on the memory of the channel, we show the true advantage of parallel systems. We continue to investigate this new finding and provide insight into when capacity should be pooled in one channel or when independent parallel channels perform better. We complement our analysis with simulation results and evaluate different update policies, scheduling policies, and the use of heterogeneous channels that is most relevant for latest multi-band networks.

This paper presents the Zurich Transit Bus (ZTBus) dataset, which consists of data recorded during driving missions of electric city buses in Zurich, Switzerland. The data was collected over several years on two trolley buses as part of multiple research projects. It involves more than a thousand missions across all seasons, each mission usually covering a full day of operation. The ZTBus dataset contains detailed information on the vehicle's power demand, propulsion system, odometry, global position, ambient temperature, door openings, number of passengers, dispatch patterns within the public transportation network, etc. All signals are synchronized in time and include an absolute timestamp in tabular form. The dataset can be used as a foundation for a variety of studies and analyses. For example, the data can serve as a basis for simulations to estimate the performance of different public transit vehicle types, or to evaluate and optimize control strategies of hybrid electric vehicles. Furthermore, numerous influencing factors on vehicle operation, such as traffic, passenger volume, etc., can be analyzed in detail., Comment: Published version

Solving tasks such as speaker recognition, music classification, or semantic audio event tagging with deep learning models typically requires computationally demanding networks. General-purpose audio embeddings (GPAEs) are dense representations of audio signals that allow lightweight, shallow classifiers to tackle various audio tasks. The idea is that a single complex feature extractor would extract dense GPAEs, while shallow MLPs can produce task-specific predictions. If the extracted dense representations are general enough to allow the simple downstream classifiers to generalize to a variety of tasks in the audio domain, a single costly forward pass suffices to solve multiple tasks in parallel. In this work, we try to reduce the cost of GPAE extractors to make them suitable for resource-constrained devices. We use efficient MobileNets trained on AudioSet using Knowledge Distillation from a Transformer ensemble as efficient GPAE extractors. We explore how to obtain high-quality GPAEs from the model, study how model complexity relates to the quality of extracted GPAEs, and conclude that low-complexity models can generate competitive GPAEs, paving the way for analyzing audio streams on edge devices w.r.t. multiple audio classification and recognition tasks., Comment: In Proceedings of the 31st European Signal Processing Conference, EUSIPCO 2023. Source Code available at: https://github.com/fschmid56/EfficientAT_HEAR

Expert musicians can mould a musical piece to convey specific emotions that they intend to communicate. In this paper, we place a mid-level features based music emotion model in this performer-to-listener communication scenario, and demonstrate via a small visualisation music emotion decoding in real time. We also extend the existing set of mid-level features using analogues of perceptual speed and perceived dynamics., Comment: Extended version of Late-Breaking Demo Session paper accepted at ISMIR 2022 (23rd Int. Society for Music Information Retrieval Conf., Bengaluru, India, 2022)

In this paper, we present a novel approach to perform highly efficient numerical simulations of the heating, ventilation, and air-conditioning (HVAC) system of an electric city bus. The models for this simulation are based on the assumption of a steady-state operation. We show two approaches to obtain the minimum energy requirement for a certain thermal comfort criterion under specific ambient conditions. Due to the computationally efficient approach developed, we can evaluate the model on a large dataset of 7500 scenarios in various ambient conditions to estimate the year-round performance of the system subject to different comfort requirements. Compared to a heating strategy based on positive temperature coefficient (PTC) elements, we can thus show that a heat pump (HP) can reduce the annual mean power consumption by up to 60%. Ceiling-mounted radiant heating elements complementing a PTC heating system can reduce the annual mean power consumption by up to 10%, while they cannot improve the energy efficiency when used in conjunction with a HP. Finally, a broad sensitivity study reveals the fact that improving the HP's heating-mode coefficient of performance (COP) manifests the largest leverage in terms of mean annual power consumption. Moreover, the annual energy expenditure for cooling are around eight times smaller than those for heating. The case study considered thus reveals that the advantages of improving the COP of the cooling mode are significantly lower.

Phillip L. Geissler made important contributions to the statistical mechanics of biological polymers, heterogeneous materials, and chemical dynamics in aqueous environments. He devised analytical and computational methods that revealed the underlying organization of complex systems at the frontiers of biology, chemistry, and materials science. In this retrospective, we celebrate his work at these frontiers.

In Oncology, trials evaluating drug combinations are becoming more common. While combination therapies bring the potential for greater efficacy, they also create unique challenges for ensuring drug safety. In Phase-I dose escalation trials of drug combinations, model-based approaches enable efficient use of information gathered, but the models need to account for trial complexities: appropriate modeling of interactions becomes increasingly important with growing numbers of drugs being tested simultaneously in a given trial. In principle, we can use data from multiple arms testing varying combinations to jointly estimate toxicity of the drug combinations. However, such efforts have highlighted limitations when modelling drug-drug interactions in the Bayesian Logistic Regression Model (BLRM) framework used to ensure patient safety. Previous models either do not account for non-monotonicity due to antagonistic toxicity, or exhibit the fundamental flaw of exponentially overpowering the contributions of the individual drugs in the dose-response. This specifically leads to issues when drug combinations exhibit antagonistic toxicity, in which case the toxicity probability gets vanishingly small as doses get very large. We put forward additional constraints inspired by Paracelsus' intuition of "the dose makes the poison" which avoid this flaw and present an improved interaction model which is compatible with these constraints. We create instructive data scenarios that showcase the improved behavior of this more constrained drug-drug interaction model in terms of preventing further dosing at overly toxic dose combinations and more sensible dose-finding under antagonistic drug toxicity. This model is now available in the open-source OncoBayes2 R package that implements the BLRM framework for an arbitrary number of drugs and trial arms., Comment: 14 pages, 3 figures

Tomorrow's massive-scale IoT sensor networks are poised to drive uplink traffic demand, especially in areas of dense deployment. To meet this demand, however, network designers leverage tools that often require accurate estimates of Channel State Information (CSI), which incurs a high overhead and thus reduces network throughput. Furthermore, the overhead generally scales with the number of clients, and so is of special concern in such massive IoT sensor networks. While prior work has used transmissions over one frequency band to predict the channel of another frequency band on the same link, this paper takes the next step in the effort to reduce CSI overhead: predict the CSI of a nearby but distinct link. We propose Cross-Link Channel Prediction (CLCP), a technique that leverages multi-view representation learning to predict the channel response of a large number of users, thereby reducing channel estimation overhead further than previously possible. CLCP's design is highly practical, exploiting channel estimates obtained from existing transmissions instead of dedicated channel sounding or extra pilot signals. We have implemented CLCP for two different Wi-Fi versions, namely 802.11n and 802.11ax, the latter being the leading candidate for future IoT networks. We evaluate CLCP in two large-scale indoor scenarios involving both line-of-sight and non-line-of-sight transmissions with up to 144 different 802.11ax users. Moreover, we measure its performance with four different channel bandwidths, from 20 MHz up to 160 MHz. Our results show that CLCP provides a 2x throughput gain over baseline 802.11ax and a 30 percent throughput gain over existing cross-band prediction algorithms., Comment: 13 pages, 12 figures

The propagation of chirality across scales is a common but poorly understood phenomenon in soft matter. In this work, we use computer simulations to study chiral monolayer assemblies formed by hard rod-like colloidal particles in the presence of non-adsorbing polymer and characterize the thermodynamic driving forces responsible for the twisting. Simulations show that straight (achiral) rods assemble into monolayers with a spontaneous twist that is either left- or right-handed, while helical (chiral) rods lead to assemblies with preferential chiral features that depend on their handedness and curliness. The onset of chirality in these monolayers can be traced back to small clusters formed at the initial stage of the self-assembly. In these microscopic monolayers, entropy drives twisting in ways that differ from the assumptions on which existing continuum theory is built. Depending on the geometry of the constituent rods, the preferred chiral twist can be driven by entropy gain of the polymers, or of the rods, or both. In addition, the variation of the polymer entropy with twist depends on changes in both the surface area and the volume of the monolayer. Rod fluctuations perpendicular to the monolayer also play an important role in stabilising the twisting., Comment: 14 pages, 12 figures

This paper describes several improvements to a new method for signal decomposition that we recently formulated under the name of Differentiable Dictionary Search (DDS). The fundamental idea of DDS is to exploit a class of powerful deep invertible density estimators called normalizing flows, to model the dictionary in a linear decomposition method such as NMF, effectively creating a bijection between the space of dictionary elements and the associated probability space, allowing a differentiable search through the dictionary space, guided by the estimated densities. As the initial formulation was a proof of concept with some practical limitations, we will present several steps towards making it scalable, hoping to improve both the computational complexity of the method and its signal decomposition capabilities. As a testbed for experimental evaluation, we choose the task of frame-level piano transcription, where the signal is to be decomposed into sources whose activity is attributed to individual piano notes. To highlight the impact of improved non-linear modelling of sources, we compare variants of our method to a linear overcomplete NMF baseline. Experimental results will show that even in the absence of additional constraints, our models produce increasingly sparse and precise decompositions, according to two pertinent evaluation measures., Comment: Published in the Proceedings of the 24th International Congress on Acoustics (ICA 2022), Gyeongju, Korea, October 24-28, 2022

We introduce a novel way to incorporate prior information into (semi-) supervised non-negative matrix factorization, which we call differentiable dictionary search. It enables general, highly flexible and principled modelling of mixtures where non-linear sources are linearly mixed. We study its behavior on an audio decomposition task, and conduct an extensive, highly controlled study of its modelling capabilities., Comment: Published in the Proceedings of the 29th European Signal Processing Conference (EUSIPCO 2021), Dublin, Ireland, August 23-27, 2021 (IEEE), 441-445

The success of supervised deep learning methods is largely due to their ability to learn relevant features from raw data. Deep Neural Networks (DNNs) trained on large-scale datasets are capable of capturing a diverse set of features, and learning a representation that can generalize onto unseen tasks and datasets that are from the same domain. Hence, these models can be used as powerful feature extractors, in combination with shallower models as classifiers, for smaller tasks and datasets where the amount of training data is insufficient for learning an end-to-end model from scratch. During the past years, Convolutional Neural Networks (CNNs) have largely been the method of choice for audio processing. However, recently attention-based transformer models have demonstrated great potential in supervised settings, outperforming CNNs. In this work, we investigate the use of audio transformers trained on large-scale datasets to learn general-purpose representations. We study how the different setups in these audio transformers affect the quality of their embeddings. We experiment with the models' time resolution, extracted embedding level, and receptive fields in order to see how they affect performance on a variety of tasks and datasets, following the HEAR 2021 NeurIPS challenge evaluation setup. Our results show that representations extracted by audio transformers outperform CNN representations. Furthermore, we will show that transformers trained on Audioset can be extremely effective representation extractors for a wide range of downstream tasks., Comment: will apear in HEAR: Holistic Evaluation of Audio Representations Proceedings of Machine Learning Research PMLR 166. Source code: https://github.com/kkoutini/passt_hear21

Audio Spectrogram Transformer models rule the field of Audio Tagging, outrunning previously dominating Convolutional Neural Networks (CNNs). Their superiority is based on the ability to scale up and exploit large-scale datasets such as AudioSet. However, Transformers are demanding in terms of model size and computational requirements compared to CNNs. We propose a training procedure for efficient CNNs based on offline Knowledge Distillation (KD) from high-performing yet complex transformers. The proposed training schema and the efficient CNN design based on MobileNetV3 results in models outperforming previous solutions in terms of parameter and computational efficiency and prediction performance. We provide models of different complexity levels, scaling from low-complexity models up to a new state-of-the-art performance of .483 mAP on AudioSet. Source Code available at: https://github.com/fschmid56/EfficientAT, Comment: In Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2023. Source Code available at: https://github.com/fschmid56/EfficientAT

There is an unmet clinical need for developing novel methods to complement and replace the current radiation-emitting imaging-based methods for the detection of loose pedicle screws as a complication after spinal fusion surgery which fail to identify a substantial amount of loose implants. In this work, we propose a new methodology and paradigm for the radiation-free, non-destructive, and easy-to-integrate detection of pedicle screw loosening based on vibroacoustic sensing. Furthermore, we propose a novel simulation technique for pedicle screw loosening, which is biomechanically validated. For the detection of a loose implant, we excite the vertebra of interest with a sine sweep vibration at the spinous process and attach a custom highly-sensitive piezo contact microphone to the screw head to capture the propagated vibration characteristics which are analyzed using a detection pipeline based on spectrogram features and a SE-ResNet-18. To validate the proposed approach, we conducted experiments using four human cadaveric lumbar spine specimens and evaluate our algorithm in a cross validation experiment. Our method reaches a sensitivity of 91.50 +- 6.58% and a specificity of 91.10 +- 2.27%. The proposed system shows great potentials for the development of alternative assessment methods for implant loosening based on vibroacoustic sensing.

Brain-machine interfaces (BMIs) are promising for motor rehabilitation and mobility augmentation. High-accuracy and low-power algorithms are required to achieve implantable BMI systems. In this paper, we propose a novel spiking neural network (SNN) decoder for implantable BMI regression tasks. The SNN is trained with enhanced spatio-temporal backpropagation to fully leverage its ability in handling temporal problems. The proposed SNN decoder achieves the same level of correlation coefficient as the state-of-the-art ANN decoder in offline finger velocity decoding tasks, while it requires only 6.8% of the computation operations and 9.4% of the memory access.

Concept induction, which is based on formal logical reasoning over description logics, has been used in ontology engineering in order to create ontology (TBox) axioms from the base data (ABox) graph. In this paper, we show that it can also be used to explain data differentials, for example in the context of Explainable AI (XAI), and we show that it can in fact be done in a way that is meaningful to a human observer. Our approach utilizes a large class hierarchy, curated from the Wikipedia category hierarchy, as background knowledge.

Reduced quasilinear (QL) and nonlinear (gradient-driven) models with scale separations, commonly used to interpret experiments and to forecast turbulent transport levels in magnetised plasmas are tested against nonlinear models without scale separations (flux-driven). Two distinct regimes of turbulence -- either far above threshold or near marginal stability -- are investigated with Boltzmann electrons. The success of reduced models especially hinges on the reproduction of nonlinear fluxes. Good agreement between models is found above threshold whilst reduced models would significantly underpredict fluxes near marginality, overlooking mesoscale flow organisation and turbulence self-advection. Constructive prescriptions whereby to improve reduced models is discussed.