Backpropagation and the brain

Is Learning in Biological Neural Networks Based on Stochastic Gradient Descent? An Analysis Using Stochastic Processes

In recent years, there has been an intense debate about how learning in biological neural networks (BNNs) differs from learning in artificial neural networks. It is often argued that the updating of connections in the brain relies only on local information, and therefore a stochastic gradient-descent type optimization method cannot be used. In this note, we study a stochastic model for supervised learning in BNNs. We show that a (continuous) gradient step occurs approximately when each learning opportunity is processed by many local updates. This result suggests that stochastic gradient descent may indeed play a role in optimizing BNNs.

Desiderata for Normative Models of Synaptic Plasticity

Normative models of synaptic plasticity use computational rationales to arrive at predictions of behavioral and network-level adaptive phenomena. In recent years, there has been an explosion of theoretical work in this realm, but experimental confirmation remains limited. In this review, we organize work on normative plasticity models in terms of a set of desiderata that, when satisfied, are designed to ensure that a given model demonstrates a clear link between plasticity and adaptive behavior, is consistent with known biological evidence about neural plasticity and yields specific testable predictions. As a prototype, we include a detailed analysis of the REINFORCE algorithm. We also discuss how new models have begun to improve on the identified criteria and suggest avenues for further development. Overall, we provide a conceptual guide to help develop neural learning theories that are precise, powerful, and experimentally testable.

Helpless infants are learning a foundation model

Humans have a protracted postnatal helplessness period, typically attributed to human-specific maternal constraints causing an early birth when the brain is highly immature. By aligning neurodevelopmental events across species, however, it has been found that humans are not born with especially immature brains compared with animal species with a shorter helpless period. Consistent with this, the rapidly growing field of infant neuroimaging has found that brain connectivity and functional activation at birth share many similarities with the mature brain. Inspired by machine learning, where deep neural networks also benefit from a 'helpless period' of pre-training, we propose that human infants are learning a foundation model: a set of fundamental representations that underpin later cognition with high performance and rapid generalisation.

Bio-inspired multimodal learning with organic neuromorphic electronics for behavioral conditioning in robotics

Biological systems interact directly with the environment and learn by receiving multimodal feedback via sensory stimuli that shape the formation of internal neuronal representations. Drawing inspiration from biological concepts such as exploration and sensory processing that eventually lead to behavioral conditioning, we present a robotic system handling objects through multimodal learning. A small-scale organic neuromorphic circuit locally integrates and adaptively processes multimodal sensory stimuli, enabling the robot to interact intelligently with its surroundings. The real-time handling of sensory stimuli via low-voltage organic neuromorphic devices with synaptic functionality forms multimodal associative connections that lead to behavioral conditioning, and thus the robot learns to avoid potentially dangerous objects. This work demonstrates that adaptive neuro-inspired circuitry with multifunctional organic materials, can accommodate locally efficient bio-inspired learning for advancing intelligent robotics.

A robust event-driven approach to always-on object recognition

We propose a neuromimetic architecture capable of always-on pattern recognition, i.e. at any time during processing. To achieve this, we have extended an existing event-based algorithm (Lagorce et al., 2017), which introduced novel spatio-temporal features as a Hierarchy Of Time-Surfaces (HOTS). Built from asynchronous events captured by a neuromorphic camera, these time surfaces allow to encode the local dynamics of a visual scene and to create an efficient event-based pattern recognition architecture. Inspired by neuroscience, we have extended this method to improve its performance. First, we add a homeostatic gain control on the activity of neurons to improve the learning of spatio-temporal patterns (Grimaldi et al., 2021). We also provide a new mathematical formalism that allows an analogy to be drawn between the HOTS algorithm and Spiking Neural Networks (SNN). Following this analogy, we transform the offline pattern categorization method into an online and event-driven layer. This classifier uses the spiking output of the network to define new time surfaces and we then perform the online classification with a neuromimetic implementation of a multinomial logistic regression. These improvements not only consistently increase the performance of the network, but also bring this event-driven pattern recognition algorithm fully online. The results have been validated on different datasets: Poker-DVS (Serrano-Gotarredona and Linares-Barranco, 2015), N-MNIST (Orchard, Jayawant et al., 2015) and DVS Gesture (Amir et al., 2017). This demonstrates the efficiency of this bio-realistic SNN for ultra-fast object recognition through an event-by-event categorization process.

Carbon emission prediction models: A review

Amidst growing concerns over the greenhouse effect, especially its consequential impacts, establishing effective Carbon Emission Prediction Models (CEPMs) to comprehend and predict CO2 emission trends is imperative for climate change mitigation. A review of 147 Carbon Emission Prediction Model (CEPM) studies revealed three predominant functions-prediction, optimization, and prediction factor selection. Statistical models, comprising 75 instances, were the most prevalent among prediction models, followed by neural network models at 21.8 %. The consistent rise in neural network model usage, particularly feedforward architectures, was observed from 2019 to 2022. A majority of CEPMs incorporated optimized approaches, with 94.4 % utilizing metaheuristic models. Parameter optimization was the primary focus, followed by structure optimization. Prediction factor selection models, employing Grey Relational Analysis (GRA) and Principal Component Analysis (PCA) for statistical and machine learning models, respectively, filtered factors effectively. Scrutinizing accuracy, pre-optimized CEPMs exhibited varied performance, Root Mean Square Error (RMSE) values spanned from 0.112 to 1635 Mt, while post-optimization led to a notable improvement, the minimum RMSE reached 0.0003 Mt, and the maximum was 95.14 Mt. Finally, we summarized the pros and cons of existing models, classified and counted the factors that influence carbon emissions, clarified the research objectives in CEPM and assessed the applied model evaluation methods and the spatial and temporal scales of existing research.

Information bottleneck-based Hebbian learning rule naturally ties working memory and synaptic updates

Deep neural feedforward networks are effective models for a wide array of problems, but training and deploying such networks presents a significant energy cost. Spiking neural networks (SNNs), which are modeled after biologically realistic neurons, offer a potential solution when deployed correctly on neuromorphic computing hardware. Still, many applications train SNNs offline, and running network training directly on neuromorphic hardware is an ongoing research problem. The primary hurdle is that back-propagation, which makes training such artificial deep networks possible, is biologically implausible. Neuroscientists are uncertain about how the brain would propagate a precise error signal backward through a network of neurons. Recent progress addresses part of this question, e.g., the weight transport problem, but a complete solution remains intangible. In contrast, novel learning rules based on the information bottleneck (IB) train each layer of a network independently, circumventing the need to propagate errors across layers. Instead, propagation is implicit due the layers' feedforward connectivity. These rules take the form of a three-factor Hebbian update a global error signal modulates local synaptic updates within each layer. Unfortunately, the global signal for a given layer requires processing multiple samples concurrently, and the brain only sees a single sample at a time. We propose a new three-factor update rule where the global signal correctly captures information across samples via an auxiliary memory network. The auxiliary network can be trained a priori independently of the dataset being used with the primary network. We demonstrate comparable performance to baselines on image classification tasks. Interestingly, unlike back-propagation-like schemes where there is no link between learning and memory, our rule presents a direct connection between working memory and synaptic updates. To the best of our knowledge, this is the first rule to make this link explicit. We explore these implications in initial experiments examining the effect of memory capacity on learning performance. Moving forward, this work suggests an alternate view of learning where each layer balances memory-informed compression against task performance. This view naturally encompasses several key aspects of neural computation, including memory, efficiency, and locality.

Toward the Emergence of Intelligent Control: Episodic Generalization and Optimization

Human cognition is unique in its ability to perform a wide range of tasks and to learn new tasks quickly. Both abilities have long been associated with the acquisition of knowledge that can generalize across tasks and the flexible use of that knowledge to execute goal-directed behavior. We investigate how this emerges in a neural network by describing and testing the Episodic Generalization and Optimization (EGO) framework. The framework consists of an episodic memory module, which rapidly learns relationships between stimuli; a semantic pathway, which more slowly learns how stimuli map to responses; and a recurrent context module, which maintains a representation of task-relevant context information, integrates this over time, and uses it both to recall context-relevant memories (in episodic memory) and to bias processing in favor of context-relevant features and responses (in the semantic pathway). We use the framework to address empirical phenomena across reinforcement learning, event segmentation, and category learning, showing in simulations that the same set of underlying mechanisms accounts for human performance in all three domains. The results demonstrate how the components of the EGO framework can efficiently learn knowledge that can be flexibly generalized across tasks, furthering our understanding of how humans can quickly learn how to perform a wide range of tasks-a capability that is fundamental to human intelligence.

A sparse quantized hopfield network for online-continual memory

An important difference between brains and deep neural networks is the way they learn. Nervous systems learn online where a stream of noisy data points are presented in a non-independent, identically distributed way. Further, synaptic plasticity in the brain depends only on information local to synapses. Deep networks, on the other hand, typically use non-local learning algorithms and are trained in an offline, non-noisy, independent, identically distributed setting. Understanding how neural networks learn under the same constraints as the brain is an open problem for neuroscience and neuromorphic computing. A standard approach to this problem has yet to be established. In this paper, we propose that discrete graphical models that learn via an online maximum a posteriori learning algorithm could provide such an approach. We implement this kind of model in a neural network called the Sparse Quantized Hopfield Network. We show our model outperforms state-of-the-art neural networks on associative memory tasks, outperforms these networks in online, continual settings, learns efficiently with noisy inputs, and is better than baselines on an episodic memory task.

Chaotic neural dynamics facilitate probabilistic computations through sampling

Cortical neurons exhibit highly variable responses over trials and time. Theoretical works posit that this variability arises potentially from chaotic network dynamics of recurrently connected neurons. Here, we demonstrate that chaotic neural dynamics, formed through synaptic learning, allow networks to perform sensory cue integration in a sampling-based implementation. We show that the emergent chaotic dynamics provide neural substrates for generating samples not only of a static variable but also of a dynamical trajectory, where generic recurrent networks acquire these abilities with a biologically plausible learning rule through trial and error. Furthermore, the networks generalize their experience in the stimulus-evoked samples to the inference without partial or all sensory information, which suggests a computational role of spontaneous activity as a representation of the priors as well as a tractable biological computation for marginal distributions. These findings suggest that chaotic neural dynamics may serve for the brain function as a Bayesian generative model.

A theory of brain-computer interface learning via low-dimensional control

A remarkable demonstration of the flexibility of mammalian motor systems is primates' ability to learn to control brain-computer interfaces (BCIs). This constitutes a completely novel motor behavior, yet primates are capable of learning to control BCIs under a wide range of conditions. BCIs with carefully calibrated decoders, for example, can be learned with only minutes to hours of practice. With a few weeks of practice, even BCIs with randomly constructed decoders can be learned. What are the biological substrates of this learning process? Here, we develop a theory based on a re-aiming strategy, whereby learning operates within a low-dimensional subspace of task-relevant inputs driving the local population of recorded neurons. Through comprehensive numerical and formal analysis, we demonstrate that this theory can provide a unifying explanation for disparate phenomena previously reported in three different BCI learning tasks, and we derive a novel experimental prediction that we verify with previously published data. By explicitly modeling the underlying neural circuitry, the theory reveals an interpretation of these phenomena in terms of biological constraints on neural activity.

Brain topology improved spiking neural network for efficient reinforcement learning of continuous control

The brain topology highly reflects the complex cognitive functions of the biological brain after million-years of evolution. Learning from these biological topologies is a smarter and easier way to achieve brain-like intelligence with features of efficiency, robustness, and flexibility. Here we proposed a brain topology-improved spiking neural network (BT-SNN) for efficient reinforcement learning. First, hundreds of biological topologies are generated and selected as subsets of the Allen mouse brain topology with the help of the Tanimoto hierarchical clustering algorithm, which has been widely used in analyzing key features of the brain connectome. Second, a few biological constraints are used to filter out three key topology candidates, including but not limited to the proportion of node functions (e.g., sensation, memory, and motor types) and network sparsity. Third, the network topology is integrated with the hybrid numerical solver-improved leaky-integrated and fire neurons. Fourth, the algorithm is then tuned with an evolutionary algorithm named adaptive random search instead of backpropagation to guide synaptic modifications without affecting raw key features of the topology. Fifth, under the test of four animal-survival-like RL tasks (i.e., dynamic controlling in Mujoco), the BT-SNN can achieve higher scores than not only counterpart SNN using random topology but also some classical ANNs (i.e., long-short-term memory and multi-layer perception). This result indicates that the research effort of incorporating biological topology and evolutionary learning rules has much in store for the future.

Forward layer-wise learning of convolutional neural networks through separation index maximizing

This paper proposes a forward layer-wise learning algorithm for CNNs in classification problems. The algorithm utilizes the Separation Index (SI) as a supervised complexity measure to evaluate and train each layer in a forward manner. The proposed method explains that gradually increasing the SI through layers reduces the input data's uncertainties and disturbances, achieving a better feature space representation. Hence, by approximating the SI with a variant of local triplet loss at each layer, a gradient-based learning algorithm is suggested to maximize it. Inspired by the NGRAD (Neural Gradient Representation by Activity Differences) hypothesis, the proposed algorithm operates in a forward manner without explicit error information from the last layer. The algorithm's performance is evaluated on image classification tasks using VGG16, VGG19, AlexNet, and LeNet architectures with CIFAR-10, CIFAR-100, Raabin-WBC, and Fashion-MNIST datasets. Additionally, the experiments are applied to text classification tasks using the DBPedia and AG's News datasets. The results demonstrate that the proposed layer-wise learning algorithm outperforms state-of-the-art methods in accuracy and time complexity.

Predictive coding with spiking neurons and feedforward gist signaling

Predictive coding (PC) is an influential theory in neuroscience, which suggests the existence of a cortical architecture that is constantly generating and updating predictive representations of sensory inputs. Owing to its hierarchical and generative nature, PC has inspired many computational models of perception in the literature. However, the biological plausibility of existing models has not been sufficiently explored due to their use of artificial neurons that approximate neural activity with firing rates in the continuous time domain and propagate signals synchronously. Therefore, we developed a spiking neural network for predictive coding (SNN-PC), in which neurons communicate using event-driven and asynchronous spikes. Adopting the hierarchical structure and Hebbian learning algorithms from previous PC neural network models, SNN-PC introduces two novel features: (1) a fast feedforward sweep from the input to higher areas, which generates a spatially reduced and abstract representation of input (i.e., a neural code for the gist of a scene) and provides a neurobiological alternative to an arbitrary choice of priors; and (2) a separation of positive and negative error-computing neurons, which counters the biological implausibility of a bi-directional error neuron with a very high baseline firing rate. After training with the MNIST handwritten digit dataset, SNN-PC developed hierarchical internal representations and was able to reconstruct samples it had not seen during training. SNN-PC suggests biologically plausible mechanisms by which the brain may perform perceptual inference and learning in an unsupervised manner. In addition, it may be used in neuromorphic applications that can utilize its energy-efficient, event-driven, local learning, and parallel information processing nature.

Deep CANALs: a deep learning approach to refining the canalization theory of psychopathology

Psychedelic therapy has seen a resurgence of interest in the last decade, with promising clinical outcomes for the treatment of a variety of psychopathologies. In response to this success, several theoretical models have been proposed to account for the positive therapeutic effects of psychedelics. One of the more prominent models is "RElaxed Beliefs Under pSychedelics," which proposes that psychedelics act therapeutically by relaxing the strength of maladaptive high-level beliefs encoded in the brain. The more recent "CANAL" model of psychopathology builds on the explanatory framework of RElaxed Beliefs Under pSychedelics by proposing that canalization (the development of overly rigid belief landscapes) may be a primary factor in psychopathology. Here, we make use of learning theory in deep neural networks to develop a series of refinements to the original CANAL model. Our primary theoretical contribution is to disambiguate two separate optimization landscapes underlying belief representation in the brain and describe the unique pathologies which can arise from the canalization of each. Along each dimension, we identify pathologies of either too much or too little canalization, implying that the construct of canalization does not have a simple linear correlation with the presentation of psychopathology. In this expanded paradigm, we demonstrate the ability to make novel predictions regarding what aspects of psychopathology may be amenable to psychedelic therapy, as well as what forms of psychedelic therapy may ultimately be most beneficial for a given individual.

The mediodorsal thalamus in executive control

Executive control, the ability to organize thoughts and action plans in real time, is a defining feature of higher cognition. Classical theories have emphasized cortical contributions to this process, but recent studies have reinvigorated interest in the role of the thalamus. Although it is well established that local thalamic damage diminishes cognitive capacity, such observations have been difficult to inform functional models. Recent progress in experimental techniques is beginning to enrich our understanding of the anatomical, physiological, and computational substrates underlying thalamic engagement in executive control. In this review, we discuss this progress and particularly focus on the mediodorsal thalamus, which regulates the activity within and across frontal cortical areas. We end with a synthesis that highlights frontal thalamocortical interactions in cognitive computations and discusses its functional implications in normal and pathological conditions.

Top-down generation of low-resolution representations improves visual perception and imagination

Perception or imagination requires top-down signals from high-level cortex to primary visual cortex (V1) to reconstruct or simulate the representations bottom-up stimulated by the seen images. Interestingly, top-down signals in V1 have lower spatial resolution than bottom-up representations. It is unclear why the brain uses low-resolution signals to reconstruct or simulate high-resolution representations. By modeling the top-down pathway of the visual system using the decoder of a variational auto-encoder (VAE), we reveal that low-resolution top-down signals can better reconstruct or simulate the information contained in the sparse activities of V1 simple cells, which facilitates perception and imagination. This advantage of low-resolution generation is related to facilitating high-level cortex to form geometry-respecting representations observed in experiments. Furthermore, we present two findings regarding this phenomenon in the context of AI-generated sketches, a style of drawings made of lines. First, we found that the quality of the generated sketches critically depends on the thickness of the lines in the sketches: thin-line sketches are harder to generate than thick-line sketches. Second, we propose a technique to generate high-quality thin-line sketches: instead of directly using original thin-line sketches, we use blurred sketches to train VAE or GAN (generative adversarial network), and then infer the thin-line sketches from the VAE- or GAN-generated blurred sketches. Collectively, our work suggests that low-resolution top-down generation is a strategy the brain uses to improve visual perception and imagination, which inspires new sketch-generation AI techniques.

Retrieving profile of photoresist with high aspect ratio and subwavelength features using optical spectroscopy and artificial neural network

Profile measurements of structures with a high aspect ratio and subwavelength features (HARSW) can be achieved using transmission electron microscopy and tilted scanning electron microscopy. Although electron microscopy can provide accurate HARSW measurements, it is laborious and destructive. In this paper, nondestructive and labor-saving methods were proposed to measure the dimensions of HARSW structures. The optical reflection spectrum, along with an artificial neural network (ANN) model, was adopted for interpolation with the simulation database to retrieve the dimensions of HARSW structures. To generate the ANN model, the experimental and simulated reflection spectra were adopted as the input and output variables for the training data, respectively. This ANN model can learn the discrepancy between simulation and experimental reflections. The finite-difference time-domain method was also adopted to calculate the simulated reflection spectra of HARSW structures with various dimensions, which can be used as a database. Once the experimental reflection of a HARSW structure with unknown dimensions was obtained, the ANN model could generate a simulation-like reflection spectrum. Linear regression was used to determine the correlation coefficients of the simulation-like reflection spectra in the database. The accurate dimensions of HARSW structures can be determined using a higher correlation coefficient. This methodology can be a prominent method for the process monitoring of HARSW structures.

An Agent-Based Model to Reproduce the Boolean Logic Behaviour of Neuronal Self-Organised Communities through Pulse Delay Modulation and Generation of Logic Gates

The human brain is arguably the most complex "machine" to ever exist. Its detailed functioning is yet to be fully understood, let alone modelled. Neurological processes have logical signal-processing and biophysical aspects, and both affect the brain's structure, functioning and adaptation. Mathematical approaches based on both information and graph theory have been extensively used in an attempt to approximate its biological functioning, along with Artificial Intelligence frameworks inspired by its logical functioning. In this article, an approach to model some aspects of the brain learning and signal processing is presented, mimicking the metastability and backpropagation found in the real brain while also accounting for neuroplasticity. Several simulations are carried out with this model to demonstrate how dynamic neuroplasticity, neural inhibition and neuron migration can reshape the brain's logical connectivity to synchronise signal processing and obtain certain target latencies. This work showcases the importance of dynamic logical and biophysical remodelling in brain plasticity. Combining mathematical (agents, graph theory, topology and backpropagation) and biomedical ingredients (metastability, neuroplasticity and migration), these preliminary results prove complex brain phenomena can be reproduced-under pertinent simplifications-via affordable computations, which can be construed as a starting point for more ambitiously accurate simulations.

Unsupervised learning of mid-level visual representations

Recently, a confluence between trends in neuroscience and machine learning has brought a renewed focus on unsupervised learning, where sensory processing systems learn to exploit the statistical structure of their inputs in the absence of explicit training targets or rewards. Sophisticated experimental approaches have enabled the investigation of the influence of sensory experience on neural self-organization and its synaptic bases. Meanwhile, novel algorithms for unsupervised and self-supervised learning have become increasingly popular both as inspiration for theories of the brain, particularly for the function of intermediate visual cortical areas, and as building blocks of real-world learning machines. Here we review some of these recent developments, placing them in historical context and highlighting some research lines that promise exciting breakthroughs in the near future.

Theoretical limits on the speed of learning inverse models explain the rate of adaptation in arm reaching tasks

An essential aspect of human motor learning is the formation of inverse models, which map desired actions to motor commands. Inverse models can be learned by adjusting parameters in neural circuits to minimize errors in the performance of motor tasks through gradient descent. However, the theory of gradient descent establishes limits on the learning speed. Specifically, the eigenvalues of the Hessian of the error surface around a minimum determine the maximum speed of learning in a task. Here, we use this theoretical framework to analyze the speed of learning in different inverse model learning architectures in a set of isometric arm-reaching tasks. We show theoretically that, in these tasks, the error surface and, thus the speed of learning, are determined by the shapes of the force manipulability ellipsoid of the arm and the distribution of targets in the task. In particular, rounder manipulability ellipsoids generate a rounder error surface, allowing for faster learning of the inverse model. Rounder target distributions have a similar effect. We tested these predictions experimentally in a quasi-isometric reaching task with a visuomotor transformation. The experimental results were consistent with our theoretical predictions. Furthermore, our analysis accounts for the speed of learning in previous experiments with incompatible and compatible virtual surgery tasks, and with visuomotor rotation tasks with different numbers of targets. By identifying aspects of a task that influence the speed of learning, our results provide theoretical principles for the design of motor tasks that allow for faster learning.

Learning what matters: Synaptic plasticity with invariance to second-order input correlations

Cortical populations of neurons develop sparse representations adapted to the statistics of the environment. To learn efficient population codes, synaptic plasticity mechanisms must differentiate relevant latent features from spurious input correlations, which are omnipresent in cortical networks. Here, we develop a theory for sparse coding and synaptic plasticity that is invariant to second-order correlations in the input. Going beyond classical Hebbian learning, our learning objective explains the functional form of observed excitatory plasticity mechanisms, showing how Hebbian long-term depression (LTD) cancels the sensitivity to second-order correlations so that receptive fields become aligned with features hidden in higher-order statistics. Invariance to second-order correlations enhances the versatility of biologically realistic learning models, supporting optimal decoding from noisy inputs and sparse population coding from spatially correlated stimuli. In a spiking model with triplet spike-timing-dependent plasticity (STDP), we show that individual neurons can learn localized oriented receptive fields, circumventing the need for input preprocessing, such as whitening, or population-level lateral inhibition. The theory advances our understanding of local unsupervised learning in cortical circuits, offers new interpretations of the Bienenstock-Cooper-Munro and triplet STDP models, and assigns a specific functional role to synaptic LTD mechanisms in pyramidal neurons.

Learning beyond sensations: How dreams organize neuronal representations

Semantic representations in higher sensory cortices form the basis for robust, yet flexible behavior. These representations are acquired over the course of development in an unsupervised fashion and continuously maintained over an organism's lifespan. Predictive processing theories propose that these representations emerge from predicting or reconstructing sensory inputs. However, brains are known to generate virtual experiences, such as during imagination and dreaming, that go beyond previously experienced inputs. Here, we suggest that virtual experiences may be just as relevant as actual sensory inputs in shaping cortical representations. In particular, we discuss two complementary learning principles that organize representations through the generation of virtual experiences. First, "adversarial dreaming" proposes that creative dreams support a cortical implementation of adversarial learning in which feedback and feedforward pathways engage in a productive game of trying to fool each other. Second, "contrastive dreaming" proposes that the invariance of neuronal representations to irrelevant factors of variation is acquired by trying to map similar virtual experiences together via a contrastive learning process. These principles are compatible with known cortical structure and dynamics and the phenomenology of sleep thus providing promising directions to explain cortical learning beyond the classical predictive processing paradigm.

Inferring neural activity before plasticity as a foundation for learning beyond backpropagation

For both humans and machines, the essence of learning is to pinpoint which components in its information processing pipeline are responsible for an error in its output, a challenge that is known as 'credit assignment'. It has long been assumed that credit assignment is best solved by backpropagation, which is also the foundation of modern machine learning. Here, we set out a fundamentally different principle on credit assignment called 'prospective configuration'. In prospective configuration, the network first infers the pattern of neural activity that should result from learning, and then the synaptic weights are modified to consolidate the change in neural activity. We demonstrate that this distinct mechanism, in contrast to backpropagation, (1) underlies learning in a well-established family of models of cortical circuits, (2) enables learning that is more efficient and effective in many contexts faced by biological organisms and (3) reproduces surprising patterns of neural activity and behavior observed in diverse human and rat learning experiments.

Plasticity in inhibitory networks improves pattern separation in early olfactory processing

Distinguishing between nectar and non-nectar odors presents a challenge for animals due to shared compounds in complex mixtures, where changing ratios often signify differences in reward. Changes in nectar production throughout the day and potentially many times within a forager's lifetime add to the complexity. The honeybee olfactory system, containing less than a 1000 of principal neurons in the early olfactory relay, the antennal lobe (AL), must learn to associate diverse volatile blends with rewards. We used a computational network model and live imaging of the honeybee's AL to explore the neural mechanisms and functions of the AL plasticity. Our findings revealed that when trained with a set of rewarded and unrewarded odors, the AL inhibitory network suppresses shared chemical compounds while enhancing responses to distinct compounds. This results in improved pattern separation and a more concise and efficient neural code. Our Ca2+ imaging data support our model's predictions. Furthermore, we applied these contrast enhancement principles to a Graph Convolutional Network (GCN) and found that similar mechanisms could enhance the performance of artificial neural networks. Our model provides insights into how plasticity at the inhibitory network level reshapes coding for efficient learning of complex odors.

Resilience and Disaster: Flexible Adaptation in the Face of Uncertain Threat

Disasters cause sweeping damage, hardship, and loss of life. In this article, we first consider the dominant psychological approach to disasters and its narrow focus on psychopathology (e.g., posttraumatic stress disorder). We then review research on a broader approach that has identified heterogeneous, highly replicable trajectories of outcome, the most common being stable mental health or resilience. We review trajectory research for different types of disasters, including the COVID-19 pandemic. Next, we consider correlates of the resilience trajectory and note their paradoxically limited ability to predict future resilient outcomes. Research using machine learning algorithms improved prediction but has not yet illuminated the mechanism behind resilient adaptation. To that end, we propose a more direct psychological explanation for resilience based on research on the motivational and mechanistic components of regulatory flexibility. Finally, we consider how future research might leverage new computational approaches to better capture regulatory flexibility in real time.

The grid-cell normative model: Unifying 'principles'

A normative model for the emergence of entorhinal grid cells in the brain's navigational system has been proposed (Sorscher et al., 2023. Neuron 111, 121-137). Using computational modeling of place-to-grid cell interactions, the authors characterized the fundamental nature of grid cells through information processing. However, the normative model does not consider certain discoveries that complement or contradict the conditions for such emergence. By briefly reviewing current evidence, we draw some implications on the interplay between place cell replay sequences and intrinsic grid cell oscillations related to the hippocampal-entorhinal navigation system that can extend the normative model.

Cognition of Time and Thinking Beyond

A common research protocol in cognitive neuroscience is to train subjects to perform deliberately designed experiments while recording brain activity, with the aim of understanding the brain mechanisms underlying cognition. However, how the results of this protocol of research can be applied in technology is seldom discussed. Here, I review the studies on time processing of the brain as examples of this research protocol, as well as two main application areas of neuroscience (neuroengineering and brain-inspired artificial intelligence). Time processing is a fundamental dimension of cognition, and time is also an indispensable dimension of any real-world signal to be processed in technology. Therefore, one may expect that the studies of time processing in cognition profoundly influence brain-related technology. Surprisingly, I found that the results from cognitive studies on timing processing are hardly helpful in solving practical problems. This awkward situation may be due to the lack of generalizability of the results of cognitive studies, which are under well-controlled laboratory conditions, to real-life situations. This lack of generalizability may be rooted in the fundamental unknowability of the world (including cognition). Overall, this paper questions and criticizes the usefulness and prospect of the abovementioned research protocol of cognitive neuroscience. I then give three suggestions for future research. First, to improve the generalizability of research, it is better to study brain activity under real-life conditions instead of in well-controlled laboratory experiments. Second, to overcome the unknowability of the world, we can engineer an easily accessible surrogate of the object under investigation, so that we can predict the behavior of the object under investigation by experimenting on the surrogate. Third, the paper calls for technology-oriented research, with the aim of technology creation instead of knowledge discovery.

Seeing Is Believing: Brain-Inspired Modular Training for Mechanistic Interpretability

We introduce Brain-Inspired Modular Training (BIMT), a method for making neural networks more modular and interpretable. Inspired by brains, BIMT embeds neurons in a geometric space and augments the loss function with a cost proportional to the length of each neuron connection. This is inspired by the idea of minimum connection cost in evolutionary biology, but we are the first the combine this idea with training neural networks with gradient descent for interpretability. We demonstrate that BIMT discovers useful modular neural networks for many simple tasks, revealing compositional structures in symbolic formulas, interpretable decision boundaries and features for classification, and mathematical structure in algorithmic datasets. Qualitatively, BIMT-trained networks have modules readily identifiable by the naked eye, but regularly trained networks seem much more complicated. Quantitatively, we use Newman's method to compute the modularity of network graphs; BIMT achieves the highest modularity for all our test problems. A promising and ambitious future direction is to apply the proposed method to understand large models for vision, language, and science.

Active Predictive Coding: A Unifying Neural Model for Active Perception, Compositional Learning, and Hierarchical Planning

There is growing interest in predictive coding as a model of how the brain learns through predictions and prediction errors. Predictive coding models have traditionally focused on sensory coding and perception. Here we introduce active predictive coding (APC) as a unifying model for perception, action, and cognition. The APC model addresses important open problems in cognitive science and AI, including (1) how we learn compositional representations (e.g., part-whole hierarchies for equivariant vision) and (2) how we solve large-scale planning problems, which are hard for traditional reinforcement learning, by composing complex state dynamics and abstract actions from simpler dynamics and primitive actions. By using hypernetworks, self-supervised learning, and reinforcement learning, APC learns hierarchical world models by combining task-invariant state transition networks and task-dependent policy networks at multiple abstraction levels. We illustrate the applicability of the APC model to active visual perception and hierarchical planning. Our results represent, to our knowledge, the first proof-of-concept demonstration of a unified approach to addressing the part-whole learning problem in vision, the nested reference frames learning problem in cognition, and the integrated state-action hierarchy learning problem in reinforcement learning.

Fixing the problems of deep neural networks will require better training data and learning algorithms

Bowers et al. argue that deep neural networks (DNNs) are poor models of biological vision because they often learn to rival human accuracy by relying on strategies that differ markedly from those of humans. We show that this problem is worsening as DNNs are becoming larger-scale and increasingly more accurate, and prescribe methods for building DNNs that can reliably model biological vision.

How deep is the brain? The shallow brain hypothesis

Deep learning and predictive coding architectures commonly assume that inference in neural networks is hierarchical. However, largely neglected in deep learning and predictive coding architectures is the neurobiological evidence that all hierarchical cortical areas, higher or lower, project to and receive signals directly from subcortical areas. Given these neuroanatomical facts, today's dominance of cortico-centric, hierarchical architectures in deep learning and predictive coding networks is highly questionable; such architectures are likely to be missing essential computational principles the brain uses. In this Perspective, we present the shallow brain hypothesis: hierarchical cortical processing is integrated with a massively parallel process to which subcortical areas substantially contribute. This shallow architecture exploits the computational capacity of cortical microcircuits and thalamo-cortical loops that are not included in typical hierarchical deep learning and predictive coding networks. We argue that the shallow brain architecture provides several critical benefits over deep hierarchical structures and a more complete depiction of how mammalian brains achieve fast and flexible computational capabilities.

Silences, spikes and bursts: Three-part knot of the neural code

When a neuron breaks silence, it can emit action potentials in a number of patterns. Some responses are so sudden and intense that electrophysiologists felt the need to single them out, labelling action potentials emitted at a particularly high frequency with a metonym - bursts. Is there more to bursts than a figure of speech? After all, sudden bouts of high-frequency firing are expected to occur whenever inputs surge. The burst coding hypothesis advances that the neural code has three syllables: silences, spikes and bursts. We review evidence supporting this ternary code in terms of devoted mechanisms for burst generation, synaptic transmission and synaptic plasticity. We also review the learning and attention theories for which such a triad is beneficial.

How well do rudimentary plasticity rules predict adult visual object learning?

A core problem in visual object learning is using a finite number of images of a new object to accurately identify that object in future, novel images. One longstanding, conceptual hypothesis asserts that this core problem is solved by adult brains through two connected mechanisms: 1) the re-representation of incoming retinal images as points in a fixed, multidimensional neural space, and 2) the optimization of linear decision boundaries in that space, via simple plasticity rules applied to a single downstream layer. Though this scheme is biologically plausible, the extent to which it explains learning behavior in humans has been unclear-in part because of a historical lack of image-computable models of the putative neural space, and in part because of a lack of measurements of human learning behaviors in difficult, naturalistic settings. Here, we addressed these gaps by 1) drawing from contemporary, image-computable models of the primate ventral visual stream to create a large set of testable learning models (n = 2,408 models), and 2) using online psychophysics to measure human learning trajectories over a varied set of tasks involving novel 3D objects (n = 371,000 trials), which we then used to develop (and publicly release) empirical benchmarks for comparing learning models to humans. We evaluated each learning model on these benchmarks, and found those based on deep, high-level representations from neural networks were surprisingly aligned with human behavior. While no tested model explained the entirety of replicable human behavior, these results establish that rudimentary plasticity rules, when combined with appropriate visual representations, have high explanatory power in predicting human behavior with respect to this core object learning problem.

Backpropagation-free training of deep physical neural networks

Recent successes in deep learning for vision and natural language processing are attributed to larger models but come with energy consumption and scalability issues. Current training of digital deep learning models primarily relies on backpropagation that is unsuitable for physical implementation. Here, we proposed a simple deep neural network architecture augmented by a physical local learning (PhyLL) algorithm, enabling supervised and unsupervised training of deep physical neural networks, without detailed knowledge of the nonlinear physical layer's properties. We trained diverse wave-based physical neural networks in vowel and image classification experiments, showcasing our approach's universality. Our method shows advantages over other hardware-aware training schemes by improving training speed, enhancing robustness, and reducing power consumption by eliminating the need for system modelling and thus decreasing digital computation.

Cellular computation and cognition

Contemporary neural network models often overlook a central biological fact about neural processing: that single neurons are themselves complex, semi-autonomous computing systems. Both the information processing and information storage abilities of actual biological neurons vastly exceed the simple weighted sum of synaptic inputs computed by the "units" in standard neural network models. Neurons are eukaryotic cells that store information not only in synapses, but also in their dendritic structure and connectivity, as well as genetic "marking" in the epigenome of each individual cell. Each neuron computes a complex nonlinear function of its inputs, roughly equivalent in processing capacity to an entire 1990s-era neural network model. Furthermore, individual cells provide the biological interface between gene expression, ongoing neural processing, and stored long-term memory traces. Neurons in all organisms have these properties, which are thus relevant to all of neuroscience and cognitive biology. Single-cell computation may also play a particular role in explaining some unusual features of human cognition. The recognition of the centrality of cellular computation to "natural computation" in brains, and of the constraints it imposes upon brain evolution, thus has important implications for the evolution of cognition, and how we study it.

Change detection in the primate auditory cortex through feedback of prediction error signals

Although cortical feedback signals are essential for modulating feedforward processing, no feedback error signal across hierarchical cortical areas has been reported. Here, we observed such a signal in the auditory cortex of awake common marmoset during an oddball paradigm to induce auditory duration mismatch negativity. Prediction errors to a deviant tone presentation were generated as offset calcium responses of layer 2/3 neurons in the rostral parabelt (RPB) of higher-order auditory cortex, while responses to non-deviant tones were strongly suppressed. Within several hundred milliseconds, the error signals propagated broadly into layer 1 of the primary auditory cortex (A1) and accumulated locally on top of incoming auditory signals. Blockade of RPB activity prevented deviance detection in A1. Optogenetic activation of RPB following tone presentation nonlinearly enhanced A1 tone response. Thus, the feedback error signal is critical for automatic detection of unpredicted stimuli in physiological auditory processing and may serve as backpropagation-like learning.

Breakdown of category-specific word representations in a brain-constrained neurocomputational model of semantic dementia

The neurobiological nature of semantic knowledge, i.e., the encoding and storage of conceptual information in the human brain, remains a poorly understood and hotly debated subject. Clinical data on semantic deficits and neuroimaging evidence from healthy individuals have suggested multiple cortical regions to be involved in the processing of meaning. These include semantic hubs (most notably, anterior temporal lobe, ATL) that take part in semantic processing in general as well as sensorimotor areas that process specific aspects/categories according to their modality. Biologically inspired neurocomputational models can help elucidate the exact roles of these regions in the functioning of the semantic system and, importantly, in its breakdown in neurological deficits. We used a neuroanatomically constrained computational model of frontotemporal cortices implicated in word acquisition and processing, and adapted it to simulate and explain the effects of semantic dementia (SD) on word processing abilities. SD is a devastating, yet insufficiently understood progressive neurodegenerative disease, characterised by semantic knowledge deterioration that is hypothesised to be specifically related to neural damage in the ATL. The behaviour of our brain-based model is in full accordance with clinical data-namely, word comprehension performance decreases as SD lesions in ATL progress, whereas word repetition abilities remain less affected. Furthermore, our model makes predictions about lesion- and category-specific effects of SD: our simulation results indicate that word processing should be more impaired for object- than for action-related words, and that degradation of white matter should produce more severe consequences than the same proportion of grey matter decay. In sum, the present results provide a neuromechanistic explanatory account of cortical-level language impairments observed during the onset and progress of semantic dementia.

Vectorized instructive signals in cortical dendrites during a brain-computer interface task

Backpropagation of error is the most widely used learning algorithm in artificial neural networks, forming the backbone of modern machine learning and artificial intelligence1,2. Backpropagation provides a solution to the credit assignment problem by vectorizing an error signal tailored to individual neurons. Recent theoretical models have suggested that neural circuits could implement backpropagation-like learning by semi-independently processing feedforward and feedback information streams in separate dendritic compartments3-7. This presents a compelling, but untested, hypothesis for how cortical circuits could solve credit assignment in the brain. We designed a neurofeedback brain-computer interface (BCI) task with an experimenter-defined reward function to evaluate the key requirements for dendrites to implement backpropagation-like learning. We trained mice to modulate the activity of two spatially intermingled populations (4 or 5 neurons each) of layer 5 pyramidal neurons in the retrosplenial cortex to rotate a visual grating towards a target orientation while we recorded GCaMP activity from somas and corresponding distal apical dendrites. We observed that the relative magnitudes of somatic versus dendritic signals could be predicted using the activity of the surrounding network and contained information about task-related variables that could serve as instructive signals, including reward and error. The signs of these putative teaching signals both depended on the causal role of individual neurons in the task and predicted changes in overall activity over the course of learning. These results provide the first biological evidence of a backpropagation-like solution to the credit assignment problem in the brain.

Neurobiological mechanisms for language, symbols and concepts: Clues from brain-constrained deep neural networks

Neural networks are successfully used to imitate and model cognitive processes. However, to provide clues about the neurobiological mechanisms enabling human cognition, these models need to mimic the structure and function of real brains. Brain-constrained networks differ from classic neural networks by implementing brain similarities at different scales, ranging from the micro- and mesoscopic levels of neuronal function, local neuronal links and circuit interaction to large-scale anatomical structure and between-area connectivity. This review shows how brain-constrained neural networks can be applied to study in silico the formation of mechanisms for symbol and concept processing and to work towards neurobiological explanations of specifically human cognitive abilities. These include verbal working memory and learning of large vocabularies of symbols, semantic binding carried by specific areas of cortex, attention focusing and modulation driven by symbol type, and the acquisition of concrete and abstract concepts partly influenced by symbols. Neuronal assembly activity in the networks is analyzed to deliver putative mechanistic correlates of higher cognitive processes and to develop candidate explanations founded in established neurobiological principles.

The combination of Hebbian and predictive plasticity learns invariant object representations in deep sensory networks

Recognition of objects from sensory stimuli is essential for survival. To that end, sensory networks in the brain must form object representations invariant to stimulus changes, such as size, orientation and context. Although Hebbian plasticity is known to shape sensory networks, it fails to create invariant object representations in computational models, raising the question of how the brain achieves such processing. In the present study, we show that combining Hebbian plasticity with a predictive form of plasticity leads to invariant representations in deep neural network models. We derive a local learning rule that generalizes to spiking neural networks and naturally accounts for several experimentally observed properties of synaptic plasticity, including metaplasticity and spike-timing-dependent plasticity. Finally, our model accurately captures neuronal selectivity changes observed in the primate inferotemporal cortex in response to altered visual experience. Thus, we provide a plausible normative theory emphasizing the importance of predictive plasticity mechanisms for successful representational learning.

Neurorobotic reinforcement learning for domains with parametrical uncertainty

Neuromorphic hardware paired with brain-inspired learning strategies have enormous potential for robot control. Explicitly, these advantages include low energy consumption, low latency, and adaptability. Therefore, developing and improving learning strategies, algorithms, and neuromorphic hardware integration in simulation is a key to moving the state-of-the-art forward. In this study, we used the neurorobotics platform (NRP) simulation framework to implement spiking reinforcement learning control for a robotic arm. We implemented a force-torque feedback-based classic object insertion task ("peg-in-hole") and controlled the robot for the first time with neuromorphic hardware in the loop. We therefore provide a solution for training the system in uncertain environmental domains by using randomized simulation parameters. This leads to policies that are robust to real-world parameter variations in the target domain, filling the sim-to-real gap.To the best of our knowledge, it is the first neuromorphic implementation of the peg-in-hole task in simulation with the neuromorphic Loihi chip in the loop, and with scripted accelerated interactive training in the Neurorobotics Platform, including randomized domains.

Evaluation of the Hierarchical Correspondence between the Human Brain and Artificial Neural Networks: A Review

Artificial neural networks (ANNs) that are heavily inspired by the human brain now achieve human-level performance across multiple task domains. ANNs have thus drawn attention in neuroscience, raising the possibility of providing a framework for understanding the information encoded in the human brain. However, the correspondence between ANNs and the brain cannot be measured directly. They differ in outputs and substrates, neurons vastly outnumber their ANN analogs (i.e., nodes), and the key algorithm responsible for most of modern ANN training (i.e., backpropagation) is likely absent from the brain. Neuroscientists have thus taken a variety of approaches to examine the similarity between the brain and ANNs at multiple levels of their information hierarchy. This review provides an overview of the currently available approaches and their limitations for evaluating brain-ANN correspondence.

Biologically plausible local synaptic learning rules robustly implement deep supervised learning

In deep neural networks, representational learning in the middle layer is essential for achieving efficient learning. However, the currently prevailing backpropagation learning rules (BP) are not necessarily biologically plausible and cannot be implemented in the brain in their current form. Therefore, to elucidate the learning rules used by the brain, it is critical to establish biologically plausible learning rules for practical memory tasks. For example, learning rules that result in a learning performance worse than that of animals observed in experimental studies may not be computations used in real brains and should be ruled out. Using numerical simulations, we developed biologically plausible learning rules to solve a task that replicates a laboratory experiment where mice learned to predict the correct reward amount. Although the extreme learning machine (ELM) and weight perturbation (WP) learning rules performed worse than the mice, the feedback alignment (FA) rule achieved a performance equal to that of BP. To obtain a more biologically plausible model, we developed a variant of FA, FA_Ex-100%, which implements direct dopamine inputs that provide error signals locally in the layer of focus, as found in the mouse entorhinal cortex. The performance of FA_Ex-100% was comparable to that of conventional BP. Finally, we tested whether FA_Ex-100% was robust against rule perturbations and biologically inevitable noise. FA_Ex-100% worked even when subjected to perturbations, presumably because it could calibrate the correct prediction error (e.g., dopaminergic signals) in the next step as a teaching signal if the perturbation created a deviation. These results suggest that simplified and biologically plausible learning rules, such as FA_Ex-100%, can robustly facilitate deep supervised learning when the error signal, possibly conveyed by dopaminergic neurons, is accurate.

Studying Psychosis Using Natural Language Generation: A Review of Emerging Opportunities

Disrupted language in psychotic disorders, such as schizophrenia, can manifest as false contents and formal deviations, often described as thought disorder. These features play a critical role in the social dysfunction associated with psychosis, but we continue to lack insights regarding how and why these symptoms develop. Natural language generation (NLG) is a field of computer science that focuses on generating human-like language for various applications. The theory that psychosis is related to the evolution of language in humans suggests that NLG systems that are sufficiently evolved to generate human-like language may also exhibit psychosis-like features. In this conceptual review, we propose using NLG systems that are at various stages of development as in silico tools to study linguistic features of psychosis. We argue that a program of in silico experimental research on the network architecture, function, learning rules, and training of NLG systems can help us understand better why thought disorder occurs in patients. This will allow us to gain a better understanding of the relationship between language and psychosis and potentially pave the way for new therapeutic approaches to address this vexing challenge.

Artificial Neuronal Devices Based on Emerging Materials: Neuronal Dynamics and Applications

Artificial neuronal devices are critical building blocks of neuromorphic computing systems and currently the subject of intense research motivated by application needs from new computing technology and more realistic brain emulation. Researchers have proposed a range of device concepts that can mimic neuronal dynamics and functions. Although the switching physics and device structures of these artificial neurons are largely different, their behaviors can be described by several neuron models in a more unified manner. In this paper, the reports of artificial neuronal devices based on emerging volatile switching materials are reviewed from the perspective of the demonstrated neuron models, with a focus on the neuronal functions implemented in these devices and the exploitation of these functions for computational and sensing applications. Furthermore, the neuroscience inspirations and engineering methods to enrich the neuronal dynamics that remain to be implemented in artificial neuronal devices and networks toward realizing the full functionalities of biological neurons are discussed.

Solving classification tasks by a receptron based on nonlinear optical speckle fields

Among several approaches to tackle the problem of energy consumption in modern computing systems, two solutions are currently investigated: one consists of artificial neural networks (ANNs) based on photonic technologies, the other is a different paradigm compared to ANNs and it is based on random networks of non-linear nanoscale junctions resulting from the assembling of nanoparticles or nanowires as substrates for neuromorphic computing. These networks show the presence of emergent complexity and collective phenomena in analogy with biological neural networks characterized by self-organization, redundancy, and non-linearity. Starting from this background, we propose and formalize a generalization of the perceptron model to describe a classification device based on a network of interacting units where the input weights are non-linearly dependent. We show that this model, called "receptron", provides substantial advantages compared to the perceptron as, for example, the solution of non-linearly separable Boolean functions with a single device. The receptron model is used as a starting point for the implementation of an all-optical device that exploits the non-linearity of optical speckle fields produced by a solid scatterer. By encoding these speckle fields we generated a large variety of target Boolean functions. We demonstrate that by properly setting the model parameters, different classes of functions with different multiplicity can be solved efficiently. The optical implementation of the receptron scheme opens the way for the fabrication of a completely new class of optical devices for neuromorphic data processing based on a very simple hardware.

A role for cortical interneurons as adversarial discriminators

The brain learns representations of sensory information from experience, but the algorithms by which it does so remain unknown. One popular theory formalizes representations as inferred factors in a generative model of sensory stimuli, meaning that learning must improve this generative model and inference procedure. This framework underlies many classic computational theories of sensory learning, such as Boltzmann machines, the Wake/Sleep algorithm, and a more recent proposal that the brain learns with an adversarial algorithm that compares waking and dreaming activity. However, in order for such theories to provide insights into the cellular mechanisms of sensory learning, they must be first linked to the cell types in the brain that mediate them. In this study, we examine whether a subtype of cortical interneurons might mediate sensory learning by serving as discriminators, a crucial component in an adversarial algorithm for representation learning. We describe how such interneurons would be characterized by a plasticity rule that switches from Hebbian plasticity during waking states to anti-Hebbian plasticity in dreaming states. Evaluating the computational advantages and disadvantages of this algorithm, we find that it excels at learning representations in networks with recurrent connections but scales poorly with network size. This limitation can be partially addressed if the network also oscillates between evoked activity and generative samples on faster timescales. Consequently, we propose that an adversarial algorithm with interneurons as discriminators is a plausible and testable strategy for sensory learning in biological systems.

Applications, functions, and accuracy of artificial intelligence in restorative dentistry: A literature review

OBJECTIVE

The applications of artificial intelligence (AI) are increasing in restorative dentistry; however, the AI performance is unclear for dental professionals. The purpose of this narrative review was to evaluate the applications, functions, and accuracy of AI in diverse aspects of restorative dentistry including caries detection, tooth preparation margin detection, tooth restoration design, metal structure casting, dental restoration/implant detection, removable partial denture design, and tooth shade determination.

OVERVIEW

An electronic search was performed on Medline/PubMed, Embase, Web of Science, Cochrane, Scopus, and Google Scholar databases. English-language articles, published from January 1, 2000, to March 1, 2022, relevant to the aforementioned aspects were selected using the key terms of artificial intelligence, machine learning, deep learning, artificial neural networks, convolutional neural networks, clustering, soft computing, automated planning, computational learning, computer vision, and automated reasoning as inclusion criteria. A manual search was also performed. Therefore, 157 articles were included, reviewed, and discussed.

CONCLUSIONS

Based on the current literature, the AI models have shown promising performance in the mentioned aspects when being compared with traditional approaches in terms of accuracy; however, as these models are still in development, more studies are required to validate their accuracy and apply them to routine clinical practice.

CLINICAL SIGNIFICANCE

AI with its specific functions has shown successful applications with acceptable accuracy in diverse aspects of restorative dentistry. The understanding of these functions may lead to novel applications with optimal accuracy for AI in restorative dentistry.