Learning prototypes in an unsupervised manner that respects the data density and topology is crucial for tasks such as clustering, representation learning, and visualization of high-dimensional datasets.In this paper, we propose a generalization of the Neural Gas algorithm to Riemannian manifolds, leveraging geodesic distances for prototype adaptation.The approach additionally generates a prototype neighborhood structure, enabling faithful approximation of both geometry and topology of data distributed on Riemannian manifolds.We demonstrate its effectiveness on real-world datasets from manifolds such as SO(n), S n ++ and Gr(n, k) and compare our approach to Riemannian versions of other related methods such as K-Means, K-Medoids and a Riemannian Self-Organizing Map.

Deep Learning models have achieved significant success, often facilitated by transfer learning. This involves using pre-trained models as a basis for new tasks. However, this practice carries the risk of propagating biases that are present in the original training data. In this study, we examine biases related to the protected attributes of "race", "age", and "gender" in several pre-trained classifiers that were trained on the widely used ImageNet dataset. Our analysis emphasizes intersectionality, exploring how interactions between these attributes influence biases. We introduce and employ a novel, model-agnostic approach to analyze biases in the representations of pre-trained deep neural networks through activation similarity-based clustering, with a focus on intersectionality. Our results suggest that, regardless of the specific model, ImageNet classifiers representations strongly reflect age information, cluster certain ethnic groups, and differentiate genders in middle-aged individuals.

The whisker-inspired tactile sensor is advantageous for enhancing robotic perception in proximate range and darkness via non-intrusive contacts. However, localizing contact along the whisker shaft is challenging due to the non-injective mapping between tangential contacts and the resulting bending moments at the whisker base. Previous studies suggest that incorporating axial force measurements can resolve this ambiguity. In this work, we develop a magnetically transduced whisker sensor that integrates axial force sensing as an additional mechanical signal. The sensor features a tapered whisker with a custom slope and a 3-DoF suspension mechanism, enabling axial displacement at the base, which is proportional to the applied axial force. We construct a Penalized Gaussian Process model trained on synthetic data to estimate the whisker’s motion and refine it with real-data constraints. The design is compact, low-cost, and validated through simulations and real-world experiments to differentiate tangential contacts. Furthermore, we propose an optimization-based approach for estimating instantaneous contact locations. Experimental results demonstrate that the proposed method can effectively track contacts in millimeter-level accuracy with a mean error of 7.17 mm, achieving a higher accuracy with only 4.02 mm in large-deflection and close-to-base regions.

Deep neural networks often exhibit sub-optimal performance under covariate and category shifts. Source-Free Domain Adaptation (SFDA) presents a promising solution to this dilemma, yet most SFDA approaches are restricted to closed-set scenarios. In this paper, we explore Source-Free Universal Domain Adaptation (SF-UniDA) aiming to accurately classify "known" data belonging to common categories and segregate them from target-private "unknown" data. We propose a novel Global and Local Clustering (GLC) technique, which comprises an adaptive one-vs-all global clustering algorithm to discern between target classes, complemented by a local k-NN clustering strategy to mitigate negative transfer. Despite the effectiveness, the inherent closed-set source architecture leads to uniform treatment of "unknown" data, impeding the identification of distinct "unknown" categories. To address this, we evolve GLC to GLC++, integrating a contrastive affinity learning strategy. We examine the superiority of GLC and GLC++ across multiple benchmarks and category shift scenarios. Remarkably, in the most challenging open-partial-set scenarios, GLC and GLC++ surpass GATE by 16.8% and 18.9% in H-score on VisDA, respectively. GLC++ enhances the novel category clustering accuracy of GLC by 4.1% in open-set scenarios on Office-Home. Furthermore, the introduced contrastive learning strategy not only enhances GLC but also significantly facilitates existing methodologies.

Universal Domain Adaptation (UniDA) targets knowledge transfer in the presence of both covariate and label shifts. Recently, Source-free Universal Domain Adaptation (SF-UniDA) has emerged to achieve UniDA without access to source data, which tends to be more practical due to data protection policies. The main challenge lies in determining whether covariate-shifted samples belong to target-private unknown categories. Existing methods tackle this either through hand-crafted thresholding or by developing time-consuming iterative clustering strategies. In this paper, we propose a new idea of LEArning Decomposition (LEAD), which decouples features into source-known and-unknown components to identify target-private data. Technically, LEAD initially leverages the or-thogonal decomposition analysis for feature decomposition. Then, LEAD builds instance-level decision boundaries to adaptively identify target-private data. Extensive experiments across various UniDA scenarios have demonstrated the effectiveness and superiority of LEAD. Notably, in the OPDA scenario on VisDA dataset, LEAD outperforms GLC by 3.5% overall H-score and reduces 75% time to derive pseudo-labeling decision boundaries. Besides, LEAD is also appealing in that it is complementary to most existing methods. The code is available at https://github.com/ispc-lab/LEAD.

Neuroplasticity, the adaptive capacity of the nervous system, spans a broad spectrum from short-term synaptic adjustments to the establishment of new neuronal connections. While artificial neural networks (ANNs) draw inspiration from the human brain, their current implementations primarily focus on synaptic weight plasticity, notably using the backpropagation algorithm (BP). Despite its machine learning prowess, BP deviates from biological realism due to imposed constraints and lacks the brain’s superior generalization capabilities. To bridge this gap, previous works have proposed a local competitive learning rule aligned with Hebbian plasticity, yet it presents challenges in sparsity and adherence to Dale’s law. This work addresses these challenges by integrating this competitive learning rule with additional mechanisms. A nonnegativity constraint is introduced to align with Dale’s law and to induce sparsity. Weight Perturbation (WP) is employed as a biologically plausible surrogate gradient. Homeostatic plasticity counters long-term potentiation (LTP) induced by Hebbian plasticity. The resulting framework offers a biologically plausible model, maintaining robust classification performance despite the constraint of a purely excitatory network. This learning rule demonstrates strong sparsification abilities, even pruning entire hidden neurons, potentially preventing overparametrization, thereby fostering the potential for better generalization.

Event cameras can record scene dynamics with high temporal resolution, providing rich scene details for monocular depth estimation (MDE) even at low-level illumination. Therefore, existing complementary learning approaches for MDE fuse intensity information from images and scene details from event data for better scene understanding. However, most methods directly fuse two modalities at pixel level, ignoring that the attractive complementarity mainly impacts high-level patterns that only occupy a few pixels. For example, event data is likely to complement contours of scene objects. In this paper, we discretize the scene into a set of high-level patterns to explore the complementarity and propose a Pattern-based Complementary learning architecture for monocular Depth estimation (PCDepth). Concretely, PCDepth comprises two primary components: a complementary visual representation learning module for discretizing the scene into high-level patterns and integrating complementary patterns across modalities and a refined depth estimator aimed at scene reconstruction and depth prediction while maintaining an efficiency-accuracy balance. Through pattern-based complementary learning, PCDepth fully exploits two modalities and achieves more accurate predictions than existing methods, especially in challenging nighttime scenarios. Extensive experiments on MVSEC and DSEC datasets verify the effectiveness and superiority of our PCDepth. Remarkably, compared with state-of-the-art, PCDepth achieves a 37.9% improvement in accuracy in MVSEC nighttime scenarios.

Effective communication between robots and human operators is crucial for seamless collaboration in industrial settings. We aim to capture the requirements of collaborative assembly procedures in the context of the Industry 5.0 paradigm and explore the feasibility of various multi-modal feedback systems, including visual cues, sound effects, and virtual eyes. To achieve this, we present a novel research platform designed to investigate human-robot interaction (HRI) strategies in collaborative assembly tasks. The platform uses LEGO bricks to simulate real-world assembly processes. An experiment is designed in which human and robot collaboratively build a structure, allowing us to investigate potential communication interfaces between them. A preliminary user study provides first insights into the perception of the robot’s visualized intentions and actions by the user. The platform setup is not intended to be a fixed system, but rather a starting point for further investigation and future studies in the field of HRI.

Deep neural networks (DNNs) often perform poorly in the presence of domain shift and category shift. How to upcycle DNNs and adapt them to the target task remains an important open problem. Unsupervised Domain Adaptation (UDA), especially recently proposed Source-free Domain Adaptation (SFDA), has become a promising technology to address this issue. Nevertheless, existing SFDA methods require that the source domain and target domain share the same label space, consequently being only applicable to the vanilla closed-set setting. In this paper, we take one step further and explore the Source-free Universal Domain Adaptation (SF-UniDA). The goal is to identify “known” data samples under both domain and category shift, and reject those “unknown” data samples (not present in source classes), with only the knowledge from standard pre-trained source model. To this end, we introduce an innovative global and local clustering learning technique (GLC). Specifically, we design a novel, adaptive one-vs-all global clustering algorithm to achieve the distinction across different target classes and introduce a local k-NN clustering strategy to alleviate negative transfer. We examine the superiority of our GLC on multiple benchmarks with different category shift scenarios, including partial-set, open-set, and open-partial-set DA. Remarkably, in the most challenging open-partial-set DA scenario, GLC outperforms UMAD by 14.8% on the VisDA benchmark. The code is available at https://github.com/ispc-lab/GLC.

Abstract Legged locomotion is a challenging task in the field of robotics but a rather simple one in nature. This motivates the use of biological methodologies as solutions to this problem. Central pattern generators are neural networks that are thought to be responsible for locomotion in humans and some animal species. As for robotics, many attempts were made to reproduce such systems and use them for a similar goal. One interesting design model is based on spiking neural networks. This model is the main focus of this work, as its contribution is not limited to engineering but also applicable to neuroscience. This paper introduces a new general framework for building central pattern generators that are task-independent, biologically plausible, and rely on learning methods. The abilities and properties of the presented approach are not only evaluated in simulation but also in a robotic experiment. The results are very promising as the used robot was able to perform stable walking at different speeds and to change speed within the same gait cycle.

Learning-based methods have demonstrated clear advantages in controlling robot tasks, such as the information fusion abilities, strong robustness, and high accuracy. Meanwhile, the on-board systems of robots have limited computation and energy resources, which are contradictory with state-of-the-art learning approaches. They are either too lightweight to solve complex problems or too heavyweight to be used for mobile applications. On the other hand, training spiking neural networks (SNNs) with biological plausibility has great potentials of performing fast computation and energy efficiency. However, the lack of effective learning rules for SNNs impedes their wide usage in mobile robot applications. This paper addresses the problem by introducing an end to end learning approach of spiking neural networks for a lane keeping vehicle. We consider the reward-modulated spike-timing-dependent-plasticity (R-STDP) as a promising solution in training SNNs, since it combines the advantages of both reinforcement learning and the well-known STDP. We test our approach in three scenarios that a Pioneer robot is controlled to keep lanes based on an SNN. Specifically, the lane information is encoded by the event data from a neuromorphic vision sensor. The SNN is constructed using R-STDP synapses in an all-to-all fashion. We demonstrate the advantages of our approach in terms of the lateral localization accuracy by comparing with other state-of-the-art learning algorithms based on SNNs.

Biometric security systems based on predefined speech sentences are extremely common nowadays, particularly in low-cost applications where the simplicity of the hardware involved is a great advantage. Audio spoofing verification is the problem of detecting whether a speech segment acquired from such a system is genuine, or whether it was synthesized or modified by a computer in order to make it sound like an authorized person. Developing countermeasures for spoofing attacks is clearly essential for having effective biometric and security systems based on audio features, all the more significant due to recent advances in generative machine learning. Nonetheless, the problem is complicated by the possible lack of knowledge on the technique(s) used to put forward the attack, so that anti-spoofing systems should be able to withstand also spoofing attacks that were not considered explicitly in the training stage. In this paper, we analyze the use of deep recurrent networks applied to this task, i.e. networks made by the successive combination of multiple feedforward and recurrent layers. These networks are routinely used in speech recognition and language identification but, to the best of our knowledge, they were never considered for this specific problem. We evaluate several architectures on the dataset released for the ASVspoof 2015 challenge last year. We show that, by working with very standard feature extraction routines and with a minimum amount of fine-tuning, the networks can already reach very promising error rates, comparable to state-of-the-art approaches, paving the way to further investigations on the problem using deep RNN models.

In this paper, a biologically inspired 3D slithering gait for a snake-like robot is designed and implemented for the purpose of target tracking. First, by balancing the forward speed and the stability of the robot, a straight slithering gait is modelled, under which the robot can march straight, fast, and stably. Then, for the purpose of steering, the straight slithering gait is modified into a biased slithering gait. The relationship between turning radius and gait parameters is analyzed by the resistive force theory. With the head composition algorithm, we investigate the orientation problem of the snake robot's head module to obtain stable visual information during the locomotion process. Finally, with the guidance of the vision sensor mounted in the head module, target tracking simulations and prototype experiments are conducted to demonstrate the practicality and effectiveness of the slithering gait in autonomous locomotion scenarios.

Hierarchical Temporal Memory (HTM) is a computational model of the neocortex that is capable of online learning to predict and detect anomalies from continuous data streams. To make HTM also available on power-constrained robot systems, we investigate the feasibility of implementing the model on SpiNNaker, a fully programmable energy-efficient neuromorphic many core system. Our contribution is twofold: First, we propose a mapping of the HTM model components to the SpiNNaker chip architecture. Second, a prototypic implementation of this mapping is successfully evaluated for different sets of model parameters.

In this contribution, we look at technology transfer in robotics. Generally, there is a delay between a science-push and a market-pull. In order of finding means to decrease this lag, we are going to look at the causes of this effect and at the means for improving technology transfer. For this purpose, we use a variety of data sources which shed light on the current situation in Asia, Europe and North-America. First, we will examine the technology-readiness level (TRL) scale which can be used as a means of measuring market readiness of innovative technology. After this we will look at what means experts find useful for technology transfer. Finally, we investigate academia-industry collaboration as one tool to increase technology transfer. We demonstrate a strong collaboration between industry and academia in North America which we see as a response to the lower numbers of robots deployed in the industry in North America compared to Asia and Europe. This is an on-going trend which occurs in parallel to a global trend in growing numbers of robots in use.

The autonomous learning of concept hierarchies is still a matter of research. Here we present a learning schema for Bayesian networks which results in a nested structure of sub- and superclass relationships. It is based on so-called parent divorcing but exploits the similarity of all nodes involved as expressed by their connectivity pattern. If the procedure is applied to simple object-property pairings a nested taxonomic hierarchy emerges. We further show how the learning procedure can be aligned with basic results from developmental psychology. For this we made a set of simulations which clearly indicate that a fixed developmental order of sensory maturation is crucial for the emerging conceptual system. The learning procedure itself is biologically plausible since it works incrementally, makes use of only local information and leads to a reduced computational effort by building a more efficient representation.

In this paper we present an approach for contour based object representation. To this end we use a curvature signal gained by a level-set segmentation method. The advantage of that curvature signal is that it generates no computational overhead as it is a byproduct of standard level-set segmentation methods. Different methods for the description of the segmented objects, so called object descriptors are presented. The object descriptors are all invariant against translation, rotation and scale of the object. Furthermore we show a sparse and memory efficient representation of the descriptors for a series of objects. Finally an approach for classification of unknown objects based on ldquomemorizedrdquo objects is proposed.

The second-order statistics of natural images can be well characterized by a "self-similar" 1/F/sup 2/ power spectrum and the bandpass decomposition in biological vision systems is characterized by a self-similar, wavelet-like structuring of the "frequency channels". It has thus often been suggested that there might exist a systematic interrelationship between these two properties, but a complete formal derivation of this relation has not yet been provided. Using rate-distortion arguments and a complexity measure, we first show that a self-similar bandpass decomposition can achieve a desired level of distortion with a less complex system structure than required for a decomposition in bands of equal linear bandwidth. A closer analysis reveals that the true optimum decomposition is approximately self-similar but shows a systematic decrease of the log-bandwidths with increasing center frequency of the subbands. Since this effect has also been observed in neurophysiological experiments, we conclude that the typical properties of visual neurons may in fact result from an optimized exploitation of the statistical redundancies of the natural environment.