Search

Next-item recommender systems are often trained using only positive feedback with randomly-sampled negative feedback. We show the benefits of using real negative feedback both as inputs into the user sequence and also as negative targets for training a next-song recommender system for internet radio. In particular, using explicit negative samples during training helps reduce training time by ~60% while also improving test accuracy by ~6%; adding user skips as additional inputs also can considerably increase user coverage alongside slightly improving accuracy. We test the impact of using a large number of random negative samples to capture a 'harder' one and find that the test accuracy increases with more randomly-sampled negatives, but only to a point. Too many random negatives leads to false negatives that limits the lift, which is still lower than if using true negative feedback. We also find that the test accuracy is fairly robust with respect to the proportion of different feedback types, and compare the learned embeddings for different feedback types., Comment: 6 pages, 4 figures, accepted to ACM UMAP 2024

Solar convection is visible as a net blueshift of absorption lines, which becomes apparent when observing quiet Sun granulation. This blueshift exhibits variations from the disc centre to the solar limb due to differing projection angles onto the solar atmosphere. Our goal is to investigate convective Doppler velocities based on observations from the disc centre to the solar limb. Consequently, we aim to improve our understanding of atmospheric hydrodynamics and contribute to the improvement of solar and stellar atmospheric models. We used resolved quiet-Sun spectra to investigate the convective velocity shifts of more than 1000\,\ion{Fe}{I} lines across multiple centre-to-limb positions on the solar disc. We determined the Doppler velocities with respect to the line depth. Additionally, we calculated the formation temperature and investigated its correlation with Doppler velocities. The general behaviour of convective line shifts shows a decreasing blueshift as the lines become deeper for all observing positions from the centre to limb. For spectra obtained at the solar limb, even deeper lines exhibit redshifts. We observe a velocity trend for the different observation angles, with a less pronounced convective blueshift towards the solar limb. Convective velocities show a wavelength dependence for each observing angle when analysing on the basis of line depths. We observe a decreasing convective blueshift as the formation temperatures of the lines decrease. The velocity change over temperature ranges proceeds slower towards the solar limb. When investigating Doppler velocities with respect to formation temperature, the disc centre does not exhibit the strongest blueshift.

Music retrieval and recommendation applications often rely on content features encoded as embeddings, which provide vector representations of items in a music dataset. Numerous complementary embeddings can be derived from processing items originally represented in several modalities, e.g., audio signals, user interaction data, or editorial data. However, data of any given modality might not be available for all items in any music dataset. In this work, we propose a method based on contrastive learning to combine embeddings from multiple modalities and explore the impact of the presence or absence of embeddings from diverse modalities in an artist similarity task. Experiments on two datasets suggest that our contrastive method outperforms single-modality embeddings and baseline algorithms for combining modalities, both in terms of artist retrieval accuracy and coverage. Improvements with respect to other methods are particularly significant for less popular query artists. We demonstrate our method successfully combines complementary information from diverse modalities, and is more robust to missing modality data (i.e., it better handles the retrieval of artists with different modality embeddings than the query artist's).

In this manuscript we describe the investigation of the SARS-CoV2 membrane fusion timescale by means of small-angle neutron scattering (SANS) using hydrogen/deuterium contrast variation. After the successful production of virus-like vesicles and human-host-cell-like vesicles we were able to follow the fusion of the respective vesicles in real-time. This was done using deuterated and protonated phospholipids in the vesicles in a neutron-contrast matched solvent. The vesicles were identical apart from either the presence or absence of the SARS-CoV2 spike protein. The human-host-cell-like vesicles were carrying an ACE2 receptor protein in all cases. In case of the absence of the spike protein a fusion over several hours was observed in agreement with literature, with a time constant of 4.5 h. In comparison, there was not time-evolution, but immediate fusion of the vesicles when the spike protein was present. Those two figures, fusion over several hours and fusion below 10 s corresponding to the absence or presence of the spike protein allow an upper-limit estimate for the fusion times of virus-like vesicles with the SARS-CoV2 spike protein of 10 s. This very fast fusion, when compared to the case without spike protein it is a factor of 2500, can also help to explain why infection with SARS-CoV2 can be so effective and fast. Studying spike protein variants using our method may explain differences in transmissibility between SARS-CoV2 strains. In addition, the model developed here can potentially be applied to any enveloped virus., Comment: 15 pages, 7 figures, 2 tables

We develop a pairing-field formalism for ab initio studies of non-relativistic two-component fermions on a $(d\!+\!1)$-dimensional spacetime lattice. More specifically, we focus on theories where the interaction between the two components can be described by the exchange of a corresponding pairing field. The introduction of a pairing field may indeed be convenient for studies of, e.g., the finite-temperature phase structure and critical behavior of, e.g., ultracold atomic Fermi gases. Moreover, such a formalism allows to directly compute the momentum and frequency dependence of the pair propagator, from which the pair-correlation function can be extracted. For a first illustration of the application of our formalism, we compute the density equation of state and the superfluid order parameter for a gas of unpolarized fermions in $(0\!+\!1)$ dimensions by employing the complex Langevin approach to surmount the sign problem., Comment: 15 pages, 6 figures, references added, remarks from review process included

Optimizing neural networks with noisy labels is a challenging task, especially if the label set contains real-world noise. Networks tend to generalize to reasonable patterns in the early training stages and overfit to specific details of noisy samples in the latter ones. We introduce Blind Knowledge Distillation - a novel teacher-student approach for learning with noisy labels by masking the ground truth related teacher output to filter out potentially corrupted knowledge and to estimate the tipping point from generalizing to overfitting. Based on this, we enable the estimation of noise in the training data with Otsus algorithm. With this estimation, we train the network with a modified weighted cross-entropy loss function. We show in our experiments that Blind Knowledge Distillation detects overfitting effectively during training and improves the detection of clean and noisy labels on the recently published CIFAR-N dataset. Code is available at GitHub., Comment: 8 pages, 4 figures, 2 tables. Submitted to the 1st Learning and Mining with Noisy Labels Challenge on IJCAI22, see http://competition.noisylabels.com/. Code is available https://github.com/TimoK93/blind_knowledge_distillation

Thermodynamic macro variables, such as the temperature or volume macro variable, can take on a continuum of allowable values, called thermodynamic macro values. Although referring to the same macro phenomena, the macro variables of Boltzmannian Statistical Mechanics (BSM) differ from thermodynamic macro variables in an important respect: within the framework of Boltzmannian Statistical Mechanics the evolution of macro values of systems with finite available phase space is invariably modelled as discontinuous, due to the method of partitioning phase space into macro regions with sharp, fixed boundaries. Conceptually, this is at odds with the continuous evolution of macro values as described by thermodynamics, as well as with the continuous evolution of the micro state assumed in BSM. This discrepancy I call the discontinuity problem (DP). I show how it arises from BSM's framework and demonstrate its consequences, in particular for the foundational project of reducing thermodynamics to BSM: thermodynamic macro values are shown to not supervene on the corresponding BSM macro values. With supervenience being a conditio sine qua non for the kind of reduction envisaged by the foundational project, the latter is in jeopardy., Comment: Forthcoming in Synthese

In this work, we provide a broad comparative analysis of strategies for pre-training audio understanding models for several tasks in the music domain, including labelling of genre, era, origin, mood, instrumentation, key, pitch, vocal characteristics, tempo and sonority. Specifically, we explore how the domain of pre-training datasets (music or generic audio) and the pre-training methodology (supervised or unsupervised) affects the adequacy of the resulting audio embeddings for downstream tasks. We show that models trained via supervised learning on large-scale expert-annotated music datasets achieve state-of-the-art performance in a wide range of music labelling tasks, each with novel content and vocabularies. This can be done in an efficient manner with models containing less than 100 million parameters that require no fine-tuning or reparameterization for downstream tasks, making this approach practical for industry-scale audio catalogs. Within the class of unsupervised learning strategies, we show that the domain of the training dataset can significantly impact the performance of representations learned by the model. We find that restricting the domain of the pre-training dataset to music allows for training with smaller batch sizes while achieving state-of-the-art in unsupervised learning -- and in some cases, supervised learning -- for music understanding. We also corroborate that, while achieving state-of-the-art performance on many tasks, supervised learning can cause models to specialize to the supervised information provided, somewhat compromising a model's generality.

Understanding the thermal behavior of additive manufacturing (AM) processes is crucial for enhancing the quality control and enabling customized process design. Most purely physics-based computational models suffer from intensive computational costs and the need of calibrating unknown parameters, thus not suitable for online control and iterative design application. Data-driven models taking advantage of the latest developed computational tools can serve as a more efficient surrogate, but they are usually trained over a large amount of simulation data and often fail to effectively use small but high-quality experimental data. In this work, we developed a hybrid physics-based data-driven thermal modeling approach of AM processes using physics-informed neural networks. Specifically, partially observed temperature data measured from an infrared camera is combined with the physics laws to predict full-field temperature history and to discover unknown material and process parameters. In the numerical and experimental examples, the effectiveness of adding auxiliary training data and using the pretrained model on training efficiency and prediction accuracy, as well as the ability to identify unknown parameters with partially observed data, are demonstrated. The results show that the hybrid thermal model can effectively identify unknown parameters and capture the full-field temperature accurately, and thus it has the potential to be used in iterative process design and real-time process control of AM.

Current multi-physics Finite Element Method (FEM) solvers are complex systems in terms of both their mathematical complexity and lines of code. This paper proposes a skeleton generic FEM solver, named MetaFEM, in total about 5,000 lines of Julia code, which translates generic input Partial Differential Equation (PDE) weak forms into corresponding GPU-accelerated simulations with a grammar similar to FEniCS or FreeFEM. Two novel approaches differentiate MetaFEM from the common solvers: (1) the FEM kernel is based on an original theory/algorithm which explicitly processes meta-expressions, as the name suggests, and (2) the symbolic engine is a rule-based Computer Algebra System (CAS), i.e., the equations are rewritten/derived according to a set of rewriting rules instead of going through completely fixed routines, supporting easy customization by developers. Example cases in thermal conduction, linear elasticity and incompressible flow are presented to demonstrate utility.

Inferring music time structures has a broad range of applications in music production, processing and analysis. Scholars have proposed various methods to analyze different aspects of time structures, such as beat, downbeat, tempo and meter. Many state-of-the-art (SOFA) methods, however, are computationally expensive. This makes them inapplicable in real-world industrial settings where the scale of the music collections can be millions. This paper proposes a new state space and a semi-Markov model for music time structure analysis. The proposed approach turns the commonly used 2D state spaces into a 1D model through a jump-back reward strategy. It reduces the state spaces size drastically. We then utilize the proposed method for causal, joint beat, downbeat, tempo, and meter tracking, and compare it against several previous methods. The proposed method delivers similar performance with the SOFA joint causal models with a much smaller state space and a more than 30 times speedup., Comment: International Conference on Acoustics, Speech and Signal Processing (ICASSP), May. 2022