Universal Representation Research Articles

Spatiotemporal implicit neural representation as a generalized traffic data learner

Spatiotemporal Traffic Data (STTD) measures the complex dynamical behaviors of the multiscale transportation system. Existing methods aim to reconstruct STTD using low-dimensional models. However, they are limited to data-specific dimensions or source-dependent patterns, restricting them from unifying representations. Here, we present a novel paradigm to address the STTD learning problem by parameterizing STTD as an implicit neural representation. To discern the underlying dynamics in low-dimensional regimes, coordinate-based neural networks that can encode high-frequency structures are employed to directly map coordinates to traffic variables. To unravel the entangled spatial–temporal interactions, the variability is decomposed into separate processes. We further enable modeling in irregular spaces such as sensor graphs using spectral embedding. Through continuous representations, our approach enables the modeling of a variety of STTD with a unified input, thereby serving as a generalized learner of the underlying traffic dynamics. It is also shown that it can learn implicit low-rank priors and smoothness regularization from the data, making it versatile for learning different dominating data patterns. We validate its effectiveness through extensive experiments in real-world scenarios, showcasing applications from corridor to network scales. Empirical results not only indicate that our model has significant superiority over conventional low-rank models, but also highlight that the versatility of the approach extends to different data domains, output resolutions, and network topologies. Comprehensive model analyses provide further insight into the inductive bias of STTD. We anticipate that this pioneering modeling perspective could lay the foundation for universal representation of STTD in various real-world tasks. PyTorch implementations of this project is publicly available at:https://github.com/tongnie/traffic_dynamics.

An universal energy-matching design and regulation method for combined cooling, heating, and power (CCHP) systems in different scenarios

In this study, the impact of the coupling matching mechanism between the systems and the users on the energy-saving characteristics of the systems was analyzed based on a universal representation and computation method to identify the energy saving potential of the systems. The physical model was detailed derived for the relationship between the load supply and user demands under different capacity design modes and operation strategies to construct a universal coupling matching diagram, and the impact of the coupling matching parameters (operational energy efficiency, load matching degree, etc) on system integration and operation in different energy use scenarios was quantitative analyzed to illustrate the inherent mechanism of energy saving characteristics. In addition, the coupling mechanism between the systems and the users from a global perspective was reconstructed, and a coupling method of capacity design factor and dual source regulation was proposed to realize the collaborative integration of centralized energy supply systems, combined cooling, heating, and power systems and load demands, as well as the active regulation of the internal cascade utilization of the systems, so as to improve the energy saving potential of the systems under all operating conditions. Finally, a case analysis and method verification were conducted using typical summer days at Kunming hotel as an example. The results show that the coupling matching relationship is the internal mechanism that affects the energy saving of the systems, and the average energy saving rate of the whole day can be increased by 7.45% through collaborative integration and active regulation.

Effective knowledge representation and utilization for sustainable collaborative learning across heterogeneous systems

AbstractThe increasingly decentralized and private nature of data in our digital society has motivated the development of collaborative intelligent systems that enable knowledge aggregation among data owners. However, collaborative learning has only been investigated in simple settings. For example, clients are often assumed to train solution models de novo, disregarding all prior expertise. The learned model is typically represented in task‐specific forms that are not generalizable to unseen, emerging scenarios. Finally, a universal model representation is enforced among collaborators, ignoring their local compute constraints or input representations. These limitations hampers the practicality of prior collaborative systems in learning scenarios with limited task data that demand constant knowledge adaptation and transfer across information silos, tasks, and learning models, as well as the utilization of prior solution expertise. Furthermore, prior collaborative learning frameworks are not sustainable on a macro scale where participants desire fairness allocation of benefits (e.g., access to the combined model) based on their costs of participation (e.g., overhead of model sharing and training synchronization, risk of information breaches, etc.). This necessitates a new perspective of collaborative learning where the server not only aggregates but also conducts valuation of the participant's contribution, and distribute aggregated information to individuals in commensurate to their contribution. To substantiate the above vision, we propose a new research agenda on developing effective and sustainable collaborative learning frameworks across heterogeneous systems, featuring three novel computational capabilities on knowledge organization: model expression, comprehension, and valuation.

Theatre as a space of empowerment and emphasis of voices

The theatre looks out from both directions. On the one hand, the artist’s (director’s) gaze determines our (spectator’s) gaze, and on the other hand, there is the question of whether the artist’s gaze is able to see through our eyes and, of course, whose eyes they are. Are they created by the theatre itself – in the name of all its direct creators and thus in its own image and according to its own wishes, or is ‘their studio’ also open to those who come to the theatre from the substance itself – from the living substance of the dramatic texts? So also to those who find it difficult to afford a theatre ticket. Is a theatre that does not look backwards, then, still credible as a universal or legitimate representation of the world and its social relations?

Timbral brightness perception investigated through multimodal interference

Brightness is among the most studied aspects of timbre perception. Psychoacoustically, sounds described as “bright” versus “dark” typically exhibit a high versus low frequency emphasis in the spectrum. However, relatively little is known about the neurocognitive mechanisms that facilitate these metaphors we listen with. Do they originate in universal magnitude representations common to more than one sensory modality? Triangulating three different interaction paradigms, we investigated using speeded classification whether intramodal, crossmodal, and amodal interference occurs when timbral brightness, as modeled by the centroid of the spectral envelope, and pitch height/visual brightness/numerical value processing are semantically congruent and incongruent. In four online experiments varying in priming strategy, onset timing, and response deadline, 189 total participants were presented with a baseline stimulus (a pitch, gray square, or numeral) then asked to quickly identify a target stimulus that is higher/lower, brighter/darker, or greater/less than the baseline after being primed with a bright or dark synthetic harmonic tone. Results suggest that timbral brightness modulates the perception of pitch and possibly visual brightness, but not numerical value. Semantically incongruent pitch height-timbral brightness shifts produced significantly slower reaction time (RT) and higher error compared to congruent pairs. In the visual task, incongruent pairings of gray squares and tones elicited slower RTs than congruent pairings (in two experiments). No interference was observed in the number comparison task. These findings shed light on the embodied and multimodal nature of experiencing timbre.

RL-Chord: CLSTM-Based Melody Harmonization Using Deep Reinforcement Learning.

Automatic music generation is the combination of artificial intelligence and art, in which melody harmonization is a significant and challenging task. However, previous recurrent neural network (RNN)-based work fails to maintain long-term dependency and neglects the guidance of music theory. In this article, we first devise a universal chord representation with a fixed small dimension, which can cover most existing chords and is easy to expand. Then a novel melody harmonization system based on reinforcement learning (RL), RL-Chord, is proposed to generate high-quality chord progressions. Specifically, a melody conditional LSTM (CLSTM) model is put forward that learns the transition and duration of chords well, based on which RL algorithms with three well-designed reward modules are combined to construct RL-Chord. We compare three widely used RL algorithms (i.e., policy gradient, Q -learning, and actor-critic algorithms) on the melody harmonization task for the first time and prove the superiority of deep Q -network (DQN). Furthermore, a style classifier is devised to fine-tune the pretrained DQN-Chord for zero-shot Chinese folk (CF) melody harmonization. Experimental results demonstrate that the proposed model can generate harmonious and fluent chord progressions for diverse melodies. Quantitatively, DQN-Chord achieves better performance than the compared methods on multiple evaluation metrics, such as chord histogram similarity (CHS), chord tonal distance (CTD), and melody-chord tonal distance (MCTD).

Some Approximate Division and Semigroup Identities for the Mittag-Leffler Function

It is known that the Mittag-Leffler (ML) function, Eα(z), a non-local extension of the Euler exponential function ez, does not enjoy the semigroup property while ez does. The purpose of this note is to show that Eα(z) does, however, for real t, s with s small enough, enjoy the approximate semigroup property, Eα(t+s)≈Eα(t)(1+Eα,α(t)/Eα(t))s. This follows from an approximation of limh→0+Eα(±(u+h)α)/Eα(±uα) , which also yields related expressions for α→1−, and is obtained from a recently proposed universal difference quotient representation for fractional derivatives. Graphical demonstrations are presented to show that the approximations are 'reasonably accurate' for 0h≤0.1, with virtually no distinction from identity for 0h≤0.01.

Chemistry and Local Environment Adaptive Representation graphs as material descriptors

Accelerating materials design and property prediction via supervised learning involves manually generating feature vectors or performing complicated atom coordinates transformation, which restricts the model to a limited set of crystal structures or makes it challenging to provide chemical insights. In this work, we devised a unique low-dimensional featurization technique known as "Chemistry and Local Environment Adaptive Representation" (CLEAR) graphs to automatically learn the properties of materials through the chemistry of the atom connections. CLEAR is an adaptive featurization method that integrates the chemistry of atoms with their atomic environment using Voronoi nearest neighbours (NN). In addition, integrating the CLEAR descriptors with explainable machine learning models unravels its potential to obtain numerous scientific insights. We applied this approach to study the stability of compositional and configurational diverse high entropy alloys (HEAs). The proposed framework provides a universal low-dimensional representation with chemistry-informed local environment descriptors, which outperforms the prediction of phases and formation energies in HEAs.

Mother Pampa in Salman Rushdie’s Victory City

Erich Neumann's The Great Mother: An Analysis of the Archetype provides a profound and comprehensive analysis of the mother archetype, delving into its deep-rooted significance in the human psyche. Neumann explores how this universal representation of motherhood represents not only life, nourishment, and protection but also encompasses broader themes of creation, sustenance, and empowerment. By examining its manifestations across diverse cultures, myths, and religions, he uncovers the far-reaching impact of this archetype on both individual and collective consciousness. Neumann's analysis of the mother archetype particularly emphasises the mother’s role in shaping human experiences, relationships, and cultural narratives. The writer examines Salman Rushdie’s novel Victory City based on these aspects of matriarchy leading to the complex interplay between the personal and collective unconscious, highlighting how the mother archetype influences understandings of femininity, masculinity, and the dynamics of power.

New experimental measurements of X-ray emissions induced by protons of several tens of MeV on elements with Z=22 to Z=79: Toward quantitative high-energy PIXE

K-shell X-ray production cross-sections measurement campaigns were carried out at the ARRONAX cyclotron with titanium, chromium, copper, molybdenum, silver and gold targets for different energies from 15 to 68 MeV. These measurements, carried out using three detectors (SDD, CdTe and HPGe), complete the database with 66 new values. Our different measurements agree with previously published data and are consistent for all three detectors, confirming the absence of experimental bias. The universal ionisation cross-section representation allows an optimisation of the data using a 3rd-order polynomial function. This function can be used for HE-PIXE quantification for elements with atomic numbers between 22 (Ti) and 79 (Au) in the studied energy region.

On Chow Rings of Quiver Moduli

Abstract We describe the point class and Todd class in the Chow ring of a moduli space of quiver representations, building on a result of Ellingsrud–Strømme. This, together with the presentation of the Chow ring by the second author, makes it possible to compute integrals on quiver moduli. To do so, we construct a canonical morphism of universal representations in great generality, and along the way point out its relation to the Kodaira–Spencer morphism. We illustrate the results by computing some invariants of some “small” Kronecker moduli spaces. We also prove that the first non-trivial (6-dimensional) Kronecker moduli space is isomorphic to the zero locus of a general section of $\mathcal{Q}^{\vee }(1)$ on $\textrm{Gr}(2,8)$.

Association between short- and medium-term exposure to air pollutants and depressive episode using comprehensive air quality index among the population in South Korea

BackgroundCurrently, air pollution is suggested as a risk factor for depressive episodes. Our study aimed to consider multiple air pollutants simultaneously, and continuously evaluate air pollutants using comprehensive air quality index (CAI) with depressive episode risk. MethodsUsing a nationally representative sample survey from South Korea between 2014 and 2020, 20,796 participants who underwent health examination and Patient Depression Questionnaire-9 were included in the study. Six air pollutants (PM10, PM2.5, O3, CO, SO2, NO2) were measured for the analysis. Every air pollutant was standardized by air quality index (AQI) and CAI was calculated for universal representation. Using logistic regression, short- and medium-term exposure by AQI and CAI with the risk of depressive episode was calculated by odds ratio and 95 % confidence interval (CI). Furthermore, consecutive measurements of CAI over 1-month time intervals were evaluated with the risk of depressive episodes. Every analysis was conducted seasonally. ResultsThere were 950 depressive episodes occurred during the survey. An increase in AQI for short-term exposure (0–30 days) showed higher risk of depressive episode in CO, while medium-term exposure (0–120 days) showed higher risk of depressive episode in CO, SO2, PM2.5, and PM10. During the cold season, the exposure to at least one abnormal CAI within 1-month intervals over 120 days was associated with a 68 % (95 % CI 1.11–2.54) increase in the risk of depressive episode. ConclusionsShort- and medium-term exposure of air pollution may be associated with an increased risk of depressive episodes, especially for cold season.

Physics-inspired multimodal machine learning for adaptive correlation fusion based rotating machinery fault diagnosis

Multimodality is a universal characteristic of multi-source monitoring data for rotating machinery. The correlation fusion of multimodal information is a general law to strengthen the cognition of fault features, and an effective way to improve the reliability and robustness of fault diagnosis methods. However, the physical connotation gaps between multimodal information hinder the construction of correlations, preventing mainstream machine learning (ML) based intelligent diagnosis methods from reliably and effectively taking advantage of multimodal fusion. To address these issues, a physics-inspired multimodal fusion convolutional neural network (PMFN) is proposed in this paper. It is the first attempt to integrate physical knowledge into ML models to bridge physical connotation gaps between multimodal fault information. Specifically, the characterization patterns of rotating machinery fault in multimodal information are embedded in the attention mechanism to focus on representative fault features with physical properties, and generate the universal representation of multimodal information. Furthermore, the cross-modal correlation fusion module is introduced to adaptively construct the correlations of multimodal information, thereby highlighting the unique feature of unimodal information and the shared representation of multimodal information. Finally, the superiority of the proposed fusion method is verified by two cases of industrial gearbox and bearing-rotor system.

TimesURL: Self-Supervised Contrastive Learning for Universal Time Series Representation Learning

Learning universal time series representations applicable to various types of downstream tasks is challenging but valuable in real applications. Recently, researchers have attempted to leverage the success of self-supervised contrastive learning (SSCL) in Computer Vision(CV) and Natural Language Processing(NLP) to tackle time series representation. Nevertheless, due to the special temporal characteristics, relying solely on empirical guidance from other domains may be ineffective for time series and difficult to adapt to multiple downstream tasks. To this end, we review three parts involved in SSCL including 1) designing augmentation methods for positive pairs, 2) constructing (hard) negative pairs, and 3) designing SSCL loss. For 1) and 2), we find that unsuitable positive and negative pair construction may introduce inappropriate inductive biases, which neither preserve temporal properties nor provide sufficient discriminative features. For 3), just exploring segment- or instance-level semantics information is not enough for learning universal representation. To remedy the above issues, we propose a novel self-supervised framework named TimesURL. Specifically, we first introduce a frequency-temporal-based augmentation to keep the temporal property unchanged. And then, we construct double Universums as a special kind of hard negative to guide better contrastive learning. Additionally, we introduce time reconstruction as a joint optimization objective with contrastive learning to capture both segment-level and instance-level information. As a result, TimesURL can learn high-quality universal representations and achieve state-of-the-art performance in 6 different downstream tasks, including short- and long-term forecasting, imputation, classification, anomaly detection and transfer learning.

Collaborative Learning across Heterogeneous Systems with Pre-Trained Models

The increasingly decentralized and private nature of data in our digital society has motivated the development of personalized, collaborative intelligent systems that enable knowledge aggregation across multiple data owners while accommodating for their data privacy and system constraints. However, collaborative learning has only been investigated in simple and limited settings: isolated task scenarios where learning begins from scratch and does not build on prior expertise; learned model is represented in task-specific forms which are not generalizable to unseen, emerging scenarios; and more often, a universal model representation is assumed across collaborators, ignoring their local compute constraints or input representations. This restricts its practicality in continual learning scenarios with limited task data, which demand continuous adaptation and knowledge transfer across different information silos, tasks, and learning models, as well as the utilization of prior solution expertises. To overcome these limitations, my research has been focused on developing effective and scalable resource-aware collaborative learning frameworks across heterogeneous systems.

PrefAce: Face-Centric Pretraining with Self-Structure Aware Distillation

Video-based facial analysis is important for autonomous agents to understand human expressions and sentiments. However, limited labeled data is available to learn effective facial representations. This paper proposes a novel self-supervised face-centric pretraining framework, called PrefAce, which learns transferable video facial representation without labels. The self-supervised learning is performed with an effective landmark-guided global-local tube distillation. Meanwhile, a novel instance-wise update FaceFeat Cache is built to enforce more discriminative and diverse representations for downstream tasks. Extensive experiments demonstrate that the proposed framework learns universal instance-aware facial representations with fine-grained landmark details from videos. The point is that it can transfer across various facial analysis tasks, e.g., Facial Attribute Recognition (FAR), Facial Expression Recognition (FER), DeepFake Detection (DFD), and Lip Synchronization (LS). Our framework also outperforms the state-of-the-art on various downstream tasks, even in low data regimes. Code is available at https://github.com/siyuan-h/PrefAce.

REALITY AND REPRESENTATION: THE NONFICTION WRITING IN JOHN UPDIKE'S "MOROCCO

“Morocco” is one of John Updike’s short stories concerning family travel s. With the attempt to blur the gap between fiction and nonfiction, Updike experiments in “Morocco” with a new form of represented reality, which reveals a middle ground between the discourse and the depicted world. Through the portrayal of real historical events by juxtaposing the overt family travel narrative and the covert Cold War narrative, the short story mixes realistic travel writing into a fictional representation of Cold War writing. The Cold War writing, especially when represented through a backdrop of baseless fears, forms the covert narrative dynamics and integrates the narrative rupture in the whole overt family travel story where the unreliable narrative techniques in the tale, such as denarration, and disnarration, break upthe narrative progression and present various possibilities from the fissured narration. It turns out that the various overt narrative possibilities are restricted by the covert Cold War narrative, thus, the variety of overt narratives in “Morocco” is always constrained by the Cold War ideology. While the travel narrative is at variance with this particular reality, the underlying Cold War ideology, nevertheless, serves as the universal representation.

Decoding phenotypic screening: A comparative analysis of image representations

Biomedical imaging techniques such as high content screening (HCS) are valuable for drug discovery, but high costs limit their use to pharmaceutical companies. To address this issue, The JUMP-CP consortium released a massive open image dataset of chemical and genetic perturbations, providing a valuable resource for deep learning research. In this work, we aim to utilize the JUMP-CP dataset to develop a universal representation model for HCS data, mainly data generated using U2OS cells and CellPainting protocol, using supervised and self-supervised learning approaches. We propose an evaluation protocol that assesses their performance on mode of action and property prediction tasks using a popular phenotypic screening dataset. Results show that the self-supervised approach that uses data from multiple consortium partners provides representation that is more robust to batch effects whilst simultaneously achieving performance on par with standard approaches. Together with other conclusions, it provides recommendations on the training strategy of a representation model for HCS images.

Few-Shot Font Generation With Weakly Supervised Localized Representations.

Automatic few-shot font generation aims to solve a well-defined, real-world problem because manual font designs are expensive and sensitive to the expertise of designers. Existing methods learn to disentangle style and content elements by developing a universal style representation for each font style. However, this approach limits the model in representing diverse local styles because it is unsuitable for complicated letter systems. For example, Chinese characters consist of a varying number of components (often called "radical") with a highly complex structure. In this paper, we propose a novel font generation method that learns localized styles, namely component-wise style representations, instead of universal styles. The proposed style representations enable synthesizing complex local details in text designs. However, learning component-wise styles solely from a few reference glyphs is infeasible when a target script has a large number of components, for example, over 200 for Chinese. To reduce the number of required reference glyphs, we represent component-wise styles by a product of component and style factors inspired by low-rank matrix factorization. Owing to the combination of strong representation and a compact factorization strategy, our method shows remarkably better few-shot font generation results (with only eight reference glyphs) than other state-of-the-art methods. Moreover, strong locality supervision was not utilized, such as the location of each component, skeleton, or strokes. The source code is available at https://github.com/clovaai/lffont.

A Universal Cognitive Bias in Word Order: Evidence From Speakers Whose Language Goes Against It

There is a long-standing debate in cognitive science surrounding the source of commonalities among languages of the world. Indeed, there are many potential explanations for such commonalities—accidents of history, common processes of language change, memory limitations, constraints on linguistic representations, and so on. Recent research has used psycholinguistic experiments to provide empirical evidence linking common linguistic patterns to specific features of human cognition, but these experiments tend to use English speakers, who in many cases have direct experience with the common patterns of interest. Here we highlight the importance of testing populations whose languages go against cross-linguistic trends. We investigate whether adult monolingual speakers of Kîîtharaka, which has an unusual way of ordering words, mirror the word-order preferences of English speakers. We find that they do, supporting the hypothesis that universal cognitive representations play a role in shaping word order.