German I. Parisi

Vincenzo Lomonaco, Lorenzo Pellegrini, Andrea Cossu et al.

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.

Nikhil Churamani, Tolga Dimlioglu, German I. Parisi et al.

Understanding human affective behaviour, especially in the dynamics of real-world settings, requires Facial Expression Recognition (FER) models to continuously adapt to individual differences in user expression, contextual attributions, and the environment. Current (deep) Machine Learning (ML)-based FER approaches pre-trained in isolation on benchmark datasets fail to capture the nuances of real-world interactions where data is available only incrementally, acquired by the agent or robot during interactions. New learning comes at the cost of previous knowledge, resulting in catastrophic forgetting. Lifelong or Continual Learning (CL), on the other hand, enables adaptability in agents by being sensitive to changing data distributions, integrating new information without interfering with previously learnt knowledge. Positing CL as an effective learning paradigm for FER, this work presents the Continual Facial Expression Recognition (ConFER) benchmark that evaluates popular CL techniques on FER tasks. It presents a comparative analysis of several CL-based approaches on popular FER datasets such as CK+, RAF-DB, and AffectNet and present strategies for a successful implementation of ConFER for Affective Computing (AC) research. CL techniques, under different learning settings, are shown to achieve state-of-the-art (SOTA) performance across several datasets, thus motivating a discussion on the benefits of applying CL principles towards human behaviour understanding, particularly from facial expressions, as well the challenges entailed.

Vincenzo Lomonaco, Lorenzo Pellegrini, Pau Rodriguez et al.

In the last few years, we have witnessed a renewed and fast-growing interest in continual learning with deep neural networks with the shared objective of making current AI systems more adaptive, efficient and autonomous. However, despite the significant and undoubted progress of the field in addressing the issue of catastrophic forgetting, benchmarking different continual learning approaches is a difficult task by itself. In fact, given the proliferation of different settings, training and evaluation protocols, metrics and nomenclature, it is often tricky to properly characterize a continual learning algorithm, relate it to other solutions and gauge its real-world applicability. The first Continual Learning in Computer Vision challenge held at CVPR in 2020 has been one of the first opportunities to evaluate different continual learning algorithms on a common hardware with a large set of shared evaluation metrics and 3 different settings based on the realistic CORe50 video benchmark. In this paper, we report the main results of the competition, which counted more than 79 teams registered, 11 finalists and 2300$ in prizes. We also summarize the winning approaches, current challenges and future research directions.

Qi She, Fan Feng, Qi Liu et al.

This report summarizes IROS 2019-Lifelong Robotic Vision Competition (Lifelong Object Recognition Challenge) with methods and results from the top $8$ finalists (out of over~$150$ teams). The competition dataset (L)ifel(O)ng (R)obotic V(IS)ion (OpenLORIS) - Object Recognition (OpenLORIS-object) is designed for driving lifelong/continual learning research and application in robotic vision domain, with everyday objects in home, office, campus, and mall scenarios. The dataset explicitly quantifies the variants of illumination, object occlusion, object size, camera-object distance/angles, and clutter information. Rules are designed to quantify the learning capability of the robotic vision system when faced with the objects appearing in the dynamic environments in the contest. Individual reports, dataset information, rules, and released source code can be found at the project homepage: "https://lifelong-robotic-vision.github.io/competition/".

German I. Parisi, Vincenzo Lomonaco

Online continual learning (OCL) refers to the ability of a system to learn over time from a continuous stream of data without having to revisit previously encountered training samples. Learning continually in a single data pass is crucial for agents and robots operating in changing environments and required to acquire, fine-tune, and transfer increasingly complex representations from non-i.i.d. input distributions. Machine learning models that address OCL must alleviate \textit{catastrophic forgetting} in which hidden representations are disrupted or completely overwritten when learning from streams of novel input. In this chapter, we summarize and discuss recent deep learning models that address OCL on sequential input through the use (and combination) of synaptic regularization, structural plasticity, and experience replay. Different implementations of replay have been proposed that alleviate catastrophic forgetting in connectionists architectures via the re-occurrence of (latent representations of) input sequences and that functionally resemble mechanisms of hippocampal replay in the mammalian brain. Empirical evidence shows that architectures endowed with experience replay typically outperform architectures without in (online) incremental learning tasks.

German I. Parisi

The robust recognition and assessment of human actions are crucial in human-robot interaction (HRI) domains. While state-of-the-art models of action perception show remarkable results in large-scale action datasets, they mostly lack the flexibility, robustness, and scalability needed to operate in natural HRI scenarios which require the continuous acquisition of sensory information as well as the classification or assessment of human body patterns in real time. In this chapter, I introduce a set of hierarchical models for the learning and recognition of actions from depth maps and RGB images through the use of neural network self-organization. A particularity of these models is the use of growing self-organizing networks that quickly adapt to non-stationary distributions and implement dedicated mechanisms for continual learning from temporally correlated input.

German I. Parisi, Christopher Kanan

Continual learning refers to the ability of a biological or artificial system to seamlessly learn from continuous streams of information while preventing catastrophic forgetting, i.e., a condition in which new incoming information strongly interferes with previously learned representations. Since it is unrealistic to provide artificial agents with all the necessary prior knowledge to effectively operate in real-world conditions, they must exhibit a rich set of learning capabilities enabling them to interact in complex environments with the aim to process and make sense of continuous streams of (often uncertain) information. While the vast majority of continual learning models are designed to alleviate catastrophic forgetting on simplified classification tasks, here we focus on continual learning for autonomous agents and robots required to operate in much more challenging experimental settings. In particular, we discuss well-established biological learning factors such as developmental and curriculum learning, transfer learning, and intrinsic motivation and their computational counterparts for modeling the progressive acquisition of increasingly complex knowledge and skills in a continual fashion.

Pablo Barros, German I. Parisi, Stefan Wermter

Recent models of emotion recognition strongly rely on supervised deep learning solutions for the distinction of general emotion expressions. However, they are not reliable when recognizing online and personalized facial expressions, e.g., for person-specific affective understanding. In this paper, we present a neural model based on a conditional adversarial autoencoder to learn how to represent and edit general emotion expressions. We then propose Grow-When-Required networks as personalized affective memories to learn individualized aspects of emotion expressions. Our model achieves state-of-the-art performance on emotion recognition when evaluated on \textit{in-the-wild} datasets. Furthermore, our experiments include ablation studies and neural visualizations in order to explain the behavior of our model.

German I. Parisi, Xu Ji, Stefan Wermter

Lifelong learning capabilities are crucial for artificial autonomous agents operating on real-world data, which is typically non-stationary and temporally correlated. In this work, we demonstrate that dynamically grown networks outperform static networks in incremental learning scenarios, even when bounded by the same amount of memory in both cases. Learning is unsupervised in our models, a condition that additionally makes training more challenging whilst increasing the realism of the study, since humans are able to learn without dense manual annotation. Our results on artificial neural networks reinforce that structural plasticity constitutes effective prevention against catastrophic forgetting in non-stationary environments, as well as empirically supporting the importance of neurogenesis in the mammalian brain.

Di Fu, Pablo Barros, German I. Parisi et al.

The efficient integration of multisensory observations is a key property of the brain that yields the robust interaction with the environment. However, artificial multisensory perception remains an open issue especially in situations of sensory uncertainty and conflicts. In this work, we extend previous studies on audio-visual (AV) conflict resolution in complex environments. In particular, we focus on quantitatively assessing the contribution of semantic congruency during an AV spatial localization task. In addition to conflicts in the spatial domain (i.e. spatially misaligned stimuli), we consider gender-specific conflicts with male and female avatars. Our results suggest that while semantically related stimuli affect the magnitude of the visual bias (perceptually shifting the location of the sound towards a semantically congruent visual cue), humans still strongly rely on environmental statistics to solve AV conflicts. Together with previously reported results, this work contributes to a better understanding of how multisensory integration and conflict resolution can be modelled in artificial agents and robots operating in real-world environments.

Francisco Cruz, German I. Parisi, Stefan Wermter

Interactive reinforcement learning (IRL) extends traditional reinforcement learning (RL) by allowing an agent to interact with parent-like trainers during a task. In this paper, we present an IRL approach using dynamic audio-visual input in terms of vocal commands and hand gestures as feedback. Our architecture integrates multi-modal information to provide robust commands from multiple sensory cues along with a confidence value indicating the trustworthiness of the feedback. The integration process also considers the case in which the two modalities convey incongruent information. Additionally, we modulate the influence of sensory-driven feedback in the IRL task using goal-oriented knowledge in terms of contextual affordances. We implement a neural network architecture to predict the effect of performed actions with different objects to avoid failed-states, i.e., states from which it is not possible to accomplish the task. In our experimental setup, we explore the interplay of multimodal feedback and task-specific affordances in a robot cleaning scenario. We compare the learning performance of the agent under four different conditions: traditional RL, multi-modal IRL, and each of these two setups with the use of contextual affordances. Our experiments show that the best performance is obtained by using audio-visual feedback with affordancemodulated IRL. The obtained results demonstrate the importance of multi-modal sensory processing integrated with goal-oriented knowledge in IRL tasks.

German I. Parisi, Jonathan Tong, Pablo Barros et al.

Our daily perceptual experience is driven by different neural mechanisms that yield multisensory interaction as the interplay between exogenous stimuli and endogenous expectations. While the interaction of multisensory cues according to their spatiotemporal properties and the formation of multisensory feature-based representations have been widely studied, the interaction of spatial-associative neural representations has received considerably less attention. In this paper, we propose a neural network architecture that models the interaction of spatial-associative representations to perform causal inference of audiovisual stimuli. We investigate the spatial alignment of exogenous audiovisual stimuli modulated by associative congruence. In the spatial layer, topographically arranged networks account for the interaction of audiovisual input in terms of population codes. In the associative layer, congruent audiovisual representations are obtained via the experience-driven development of feature-based associations. Levels of congruency are obtained as a by-product of the neurodynamics of self-organizing networks, where the amount of neural activation triggered by the input can be expressed via a nonlinear distance function. Our novel proposal is that activity-driven levels of congruency can be used as top-down modulatory projections to spatially distributed representations of sensory input, e.g. semantically related audiovisual pairs will yield a higher level of integration than unrelated pairs. Furthermore, levels of neural response in unimodal layers may be seen as sensory reliability for the dynamic weighting of crossmodal cues. We describe a series of planned experiments to validate our model in the tasks of multisensory interaction on the basis of semantic congruence and unimodal cue reliability.

German I. Parisi, Jun Tani, Cornelius Weber et al.

Artificial autonomous agents and robots interacting in complex environments are required to continually acquire and fine-tune knowledge over sustained periods of time. The ability to learn from continuous streams of information is referred to as lifelong learning and represents a long-standing challenge for neural network models due to catastrophic forgetting. Computational models of lifelong learning typically alleviate catastrophic forgetting in experimental scenarios with given datasets of static images and limited complexity, thereby differing significantly from the conditions artificial agents are exposed to. In more natural settings, sequential information may become progressively available over time and access to previous experience may be restricted. In this paper, we propose a dual-memory self-organizing architecture for lifelong learning scenarios. The architecture comprises two growing recurrent networks with the complementary tasks of learning object instances (episodic memory) and categories (semantic memory). Both growing networks can expand in response to novel sensory experience: the episodic memory learns fine-grained spatiotemporal representations of object instances in an unsupervised fashion while the semantic memory uses task-relevant signals to regulate structural plasticity levels and develop more compact representations from episodic experience. For the consolidation of knowledge in the absence of external sensory input, the episodic memory periodically replays trajectories of neural reactivations. We evaluate the proposed model on the CORe50 benchmark dataset for continuous object recognition, showing that we significantly outperform current methods of lifelong learning in three different incremental learning scenarios

German I. Parisi, Pablo Barros, Di Fu et al.

Crossmodal conflict resolution is crucial for robot sensorimotor coupling through the interaction with the environment, yielding swift and robust behaviour also in noisy conditions. In this paper, we propose a neurorobotic experiment in which an iCub robot exhibits human-like responses in a complex crossmodal environment. To better understand how humans deal with multisensory conflicts, we conducted a behavioural study exposing 33 subjects to congruent and incongruent dynamic audio-visual cues. In contrast to previous studies using simplified stimuli, we designed a scenario with four animated avatars and observed that the magnitude and extension of the visual bias are related to the semantics embedded in the scene, i.e., visual cues that are congruent with environmental statistics (moving lips and vocalization) induce the strongest bias. We implement a deep learning model that processes stereophonic sound, facial features, and body motion to trigger a discrete behavioural response. After training the model, we exposed the iCub to the same experimental conditions as the human subjects, showing that the robot can replicate similar responses in real time. Our interdisciplinary work provides important insights into how crossmodal conflict resolution can be modelled in robots and introduces future research directions for the efficient combination of sensory observations with internally generated knowledge and expectations.

German I. Parisi, Ronald Kemker, Jose L. Part et al.

Humans and animals have the ability to continually acquire, fine-tune, and transfer knowledge and skills throughout their lifespan. This ability, referred to as lifelong learning, is mediated by a rich set of neurocognitive mechanisms that together contribute to the development and specialization of our sensorimotor skills as well as to long-term memory consolidation and retrieval. Consequently, lifelong learning capabilities are crucial for autonomous agents interacting in the real world and processing continuous streams of information. However, lifelong learning remains a long-standing challenge for machine learning and neural network models since the continual acquisition of incrementally available information from non-stationary data distributions generally leads to catastrophic forgetting or interference. This limitation represents a major drawback for state-of-the-art deep neural network models that typically learn representations from stationary batches of training data, thus without accounting for situations in which information becomes incrementally available over time. In this review, we critically summarize the main challenges linked to lifelong learning for artificial learning systems and compare existing neural network approaches that alleviate, to different extents, catastrophic forgetting. We discuss well-established and emerging research motivated by lifelong learning factors in biological systems such as structural plasticity, memory replay, curriculum and transfer learning, intrinsic motivation, and multisensory integration.

Jonathan Tong, German I. Parisi, Stefan Wermter et al.

When the brain receives input from multiple sensory systems, it is faced with the question of whether it is appropriate to process the inputs in combination, as if they originated from the same event, or separately, as if they originated from distinct events. Furthermore, it must also have a mechanism through which it can keep sensory inputs calibrated to maintain the accuracy of its internal representations. We have developed a neural network architecture capable of i) approximating optimal multisensory spatial integration, based on Bayesian causal inference, and ii) recalibrating the spatial encoding of sensory systems. The architecture is based on features of the dorsal processing hierarchy, including the spatial tuning properties of unisensory neurons and the convergence of different sensory inputs onto multisensory neurons. Furthermore, we propose that these unisensory and multisensory neurons play dual roles in i) encoding spatial location as separate or integrated estimates and ii) accumulating evidence for the independence or relatedness of multisensory stimuli. We further propose that top-down feedback connections spanning the dorsal pathway play key a role in recalibrating spatial encoding at the level of early unisensory cortices. Our proposed architecture provides possible explanations for a number of human electrophysiological and neuroimaging results and generates testable predictions linking neurophysiology with behaviour.

Pablo Barros, German I. Parisi, Di Fu et al.

The human brain is able to learn, generalize, and predict crossmodal stimuli. Learning by expectation fine-tunes crossmodal processing at different levels, thus enhancing our power of generalization and adaptation in highly dynamic environments. In this paper, we propose a deep neural architecture trained by using expectation learning accounting for unsupervised learning tasks. Our learning model exhibits a self-adaptable behavior, setting the first steps towards the development of deep learning architectures for crossmodal stimuli association.

Luiza Mici, German I. Parisi, Stefan Wermter

During visuomotor tasks, robots must compensate for temporal delays inherent in their sensorimotor processing systems. Delay compensation becomes crucial in a dynamic environment where the visual input is constantly changing, e.g., during the interacting with a human demonstrator. For this purpose, the robot must be equipped with a prediction mechanism for using the acquired perceptual experience to estimate possible future motor commands. In this paper, we present a novel neural network architecture that learns prototypical visuomotor representations and provides reliable predictions on the basis of the visual input. These predictions are used to compensate for the delayed motor behavior in an online manner. We investigate the performance of our method with a set of experiments comprising a humanoid robot that has to learn and generate visually perceived arm motion trajectories. We evaluate the accuracy in terms of mean prediction error and analyze the response of the network to novel movement demonstrations. Additionally, we report experiments with incomplete data sequences, showing the robustness of the proposed architecture in the case of a noisy and faulty visual sensor.

Luiza Mici, German I. Parisi, Stefan Wermter

The visual recognition of transitive actions comprising human-object interactions is a key component for artificial systems operating in natural environments. This challenging task requires jointly the recognition of articulated body actions as well as the extraction of semantic elements from the scene such as the identity of the manipulated objects. In this paper, we present a self-organizing neural network for the recognition of human-object interactions from RGB-D videos. Our model consists of a hierarchy of Grow-When-Required (GWR) networks that learn prototypical representations of body motion patterns and objects, accounting for the development of action-object mappings in an unsupervised fashion. We report experimental results on a dataset of daily activities collected for the purpose of this study as well as on a publicly available benchmark dataset. In line with neurophysiological studies, our self-organizing architecture exhibits higher neural activation for congruent action-object pairs learned during training sessions with respect to synthetically created incongruent ones. We show that our unsupervised model shows competitive classification results on the benchmark dataset with respect to strictly supervised approaches.