A neural data structure for novelty detection

Hybrid neural networks in the mushroom body drive olfactory preference in Drosophila

In Drosophila melanogaster, olfactory encoding in the mushroom body (MB) involves thousands of Kenyon cells (KCs) processing inputs from hundreds of projection neurons (PNs). Recent data challenge the notion of random PN-to-KC connectivity, revealing preferential connections between food-related PNs and specific KCs. Our study further uncovers a broader picture-an L-shaped hybrid network, supported by spatial patterning: Food-related PNs diverge across KC classes, whereas pheromone-sensitive PNs converge on γ KCs. α/β KCs specialize in food odors, whereas γ KCs integrate diverse inputs. Such spatial arrangement extends further to the antennal lobe (AL) and lateral horn (LH), shaping a systematic olfactory landscape. Moreover, our functional validations align with computational predictions of KC odor encoding based on the hybrid connectivity, correlating PN-KC activity with behavioral preferences. In addition, our simulations showcase the network's augmented sensitivity and precise discrimination abilities, underscoring the computational benefits of this hybrid architecture in olfactory processing.

A non-Hebbian code for episodic memory

Hebbian plasticity has long dominated neurobiological models of memory formation. Yet, plasticity rules operating on one-shot episodic memory timescales rarely depend on both pre- and postsynaptic spiking, challenging Hebbian theory in this crucial regime. Here, we present an episodic memory model governed by a simpler rule depending only on presynaptic activity. We show that this rule, capitalizing on high-dimensional neural activity with restricted transitions, naturally stores episodes as paths through complex state spaces like those underlying a world model. The resulting memory traces, which we term path vectors, are highly expressive and decodable with an odor-tracking algorithm. We show that path vectors are robust alternatives to Hebbian traces, support one-shot sequential and associative recall, along with policy learning, and shed light on specific hippocampal plasticity rules. Thus, non-Hebbian plasticity is sufficient for flexible memory and learning and well-suited to encode episodes and policies as paths through a world model.

A Drosophila-inspired intelligent olfactory biomimetic sensing system for gas recognition in complex environments

The olfactory sensory system of Drosophila has several advantages, including low power consumption, high rapidity and high accuracy. Here, we present a biomimetic intelligent olfactory sensing system based on the integration of an 18-channel microelectromechanical system (MEMS) sensor array (16 gas sensors, 1 humidity sensor and 1 temperature sensor), a complementary metal‒oxide‒semiconductor (CMOS) circuit and an olfactory lightweight machine-learning algorithm inspired by Drosophila. This system is an artificial version of the biological olfactory perception system with the capabilities of environmental sensing, multi-signal processing, and odor recognition. The olfactory data are processed and reconstructed by the combination of a shallow neural network and a residual neural network, with the aim to determine the noxious gas information in challenging environments such as high humidity scenarios and partially damaged sensor units. As a result, our electronic olfactory sensing system is capable of achieving comprehensive gas recognition by qualitatively identifying 7 types of gases with an accuracy of 98.5%, reducing the number of parameters and the difficulty of calculation, and quantitatively predicting each gas of 3-5 concentration gradients with an accuracy of 93.2%; thus, these results show superiority of our system in supporting alarm systems in emergency rescue scenarios.

Sensory encoding and memory in the mushroom body: signals, noise, and variability

To survive in changing environments, animals need to learn to associate specific sensory stimuli with positive or negative valence. How do they form stimulus-specific memories to distinguish between positively/negatively associated stimuli and other irrelevant stimuli? Solving this task is one of the functions of the mushroom body, the associative memory center in insect brains. Here we summarize recent work on sensory encoding and memory in the Drosophila mushroom body, highlighting general principles such as pattern separation, sparse coding, noise and variability, coincidence detection, and spatially localized neuromodulation, and placing the mushroom body in comparative perspective with mammalian memory systems.

Curiosity: primate neural circuits for novelty and information seeking

For many years, neuroscientists have investigated the behavioural, computational and neurobiological mechanisms that support value-based decisions, revealing how humans and animals make choices to obtain rewards. However, many decisions are influenced by factors other than the value of physical rewards or second-order reinforcers (such as money). For instance, animals (including humans) frequently explore novel objects that have no intrinsic value solely because they are novel and they exhibit the desire to gain information to reduce their uncertainties about the future, even if this information cannot lead to reward or assist them in accomplishing upcoming tasks. In this Review, I discuss how circuits in the primate brain responsible for detecting, predicting and assessing novelty and uncertainty regulate behaviour and give rise to these behavioural components of curiosity. I also briefly discuss how curiosity-related behaviours arise during postnatal development and point out some important reasons for the persistence of curiosity across generations.

Bloom filters for molecules

Ultra-large chemical libraries are reaching 10s to 100s of billions of molecules. A challenge for these libraries is to efficiently check if a proposed molecule is present. Here we propose and study Bloom filters for testing if a molecule is present in a set using either string or fingerprint representations. Bloom filters are small enough to hold billions of molecules in just a few GB of memory and check membership in sub milliseconds. We found string representations can have a false positive rate below 1% and require significantly less storage than using fingerprints. Canonical SMILES with Bloom filters with the simple FNV (Fowler-Noll-Voll) hashing function provide fast and accurate membership tests with small memory requirements. We provide a general implementation and specific filters for detecting if a molecule is purchasable, patented, or a natural product according to existing databases at https://github.com/whitead/molbloom .

Reducing Catastrophic Forgetting With Associative Learning: A Lesson From Fruit Flies

Catastrophic forgetting remains an outstanding challenge in continual learning. Recently, methods inspired by the brain, such as continual representation learning and memory replay, have been used to combat catastrophic forgetting. Associative learning (retaining associations between inputs and outputs, even after good representations are learned) plays an important function in the brain; however, its role in continual learning has not been carefully studied. Here, we identified a two-layer neural circuit in the fruit fly olfactory system that performs continual associative learning between odors and their associated valences. In the first layer, inputs (odors) are encoded using sparse, high-dimensional representations, which reduces memory interference by activating nonoverlapping populations of neurons for different odors. In the second layer, only the synapses between odor-activated neurons and the odor's associated output neuron are modified during learning; the rest of the weights are frozen to prevent unrelated memories from being overwritten. We prove theoretically that these two perceptron-like layers help reduce catastrophic forgetting compared to the original perceptron algorithm, under continual learning. We then show empirically on benchmark data sets that this simple and lightweight architecture outperforms other popular neural-inspired algorithms when also using a two-layer feedforward architecture. Overall, fruit flies evolved an efficient continual associative learning algorithm, and circuit mechanisms from neuroscience can be translated to improve machine computation.

Distributed algorithms from arboreal ants for the shortest path problem

Colonies of the arboreal turtle ant create networks of trails that link nests and food sources on the graph formed by branches and vines in the canopy of the tropical forest. Ants put down a volatile pheromone on the edges as they traverse them. At each vertex, the next edge to traverse is chosen using a decision rule based on the current pheromone level. There is a bidirectional flow of ants around the network. In a previous field study, it was observed that the trail networks approximately minimize the number of vertices, thus solving a variant of the popular shortest path problem without any central control and with minimal computational resources. We propose a biologically plausible model, based on a variant of the reinforced random walk on a graph, which explains this observation and suggests surprising algorithms for the shortest path problem and its variants. Through simulations and analysis, we show that when the rate of flow of ants does not change, the dynamics converges to the path with the minimum number of vertices, as observed in the field. The dynamics converges to the shortest path when the rate of flow increases with time, so the colony can solve the shortest path problem merely by increasing the flow rate. We also show that to guarantee convergence to the shortest path, bidirectional flow and a decision rule dividing the flow in proportion to the pheromone level are necessary, but convergence to approximately short paths is possible with other decision rules.

OneFlow: One-Class Flow for Anomaly Detection Based on a Minimal Volume Region

We propose OneFlow - a flow-based one-class classifier for anomaly (outlier) detection that finds a minimal volume bounding region. Contrary to density-based methods, OneFlow is constructed in such a way that its result typically does not depend on the structure of outliers. This is caused by the fact that during training the gradient of the cost function is propagated only over the points located near to the decision boundary (behavior similar to the support vectors in SVM). The combination of flow models and a Bernstein quantile estimator allows OneFlow to find a parametric form of bounding region, which can be useful in various applications including describing shapes from 3D point clouds. Experiments show that the proposed model outperforms related methods on real-world anomaly detection problems.

A neural theory for counting memories

Keeping track of the number of times different stimuli have been experienced is a critical computation for behavior. Here, we propose a theoretical two-layer neural circuit that stores counts of stimulus occurrence frequencies. This circuit implements a data structure, called a count sketch, that is commonly used in computer science to maintain item frequencies in streaming data. Our first model implements a count sketch using Hebbian synapses and outputs stimulus-specific frequencies. Our second model uses anti-Hebbian plasticity and only tracks frequencies within four count categories ("1-2-3-many"), which trades-off the number of categories that need to be distinguished with the potential ethological value of those categories. We show how both models can robustly track stimulus occurrence frequencies, thus expanding the traditional novelty-familiarity memory axis from binary to discrete with more than two possible values. Finally, we show that an implementation of the "1-2-3-many" count sketch exists in the insect mushroom body.

Mosquito Olfactory Response Ensemble enables pattern discovery by curating a behavioral and electrophysiological response database

Many experimental studies have examined behavioral and electrophysiological responses of mosquitoes to odors. However, the differences across studies in data collection, processing, and reporting make it difficult to perform large-scale analyses combining data from multiple studies. Here we extract and standardize data for 12 mosquito species, along with Drosophila melanogaster for comparison, from over 170 studies and curate the Mosquito Olfactory Response Ensemble (MORE), publicly available at https://neuralsystems.github.io/MORE. We demonstrate the ability of MORE in generating biological insights by finding patterns across studies. Our analyses reveal that ORs are tuned to specific ranges of several physicochemical properties of odorants; the empty-neuron recording technique for measuring OR responses is more sensitive than the Xenopus oocyte technique; there are systematic differences in the behavioral preferences reported by different types of assays; and odorants tend to become less attractive or more aversive at higher concentrations.

•

MORE is a database of behavioral and electrophysiological responses to odors

•

MORE includes data from 170 studies covering 12 species of mosquitoes along with flies

•

MORE shows differences in odor preferences measured with different assays

•

Empty-neuron technique measures responses more sensitively than the oocyte technique

A feedback control principle common to several biological and engineered systems

Feedback control is used by many distributed systems to optimize behaviour. Traditional feedback control algorithms spend significant resources to constantly sense and stabilize a continuous control variable of interest, such as vehicle speed for implementing cruise control, or body temperature for maintaining homeostasis. By contrast, discrete-event feedback (e.g. a server acknowledging when data are successfully transmitted, or a brief antennal interaction when an ant returns to the nest after successful foraging) can reduce costs associated with monitoring a continuous variable; however, optimizing behaviour in this setting requires alternative strategies. Here, we studied parallels between discrete-event feedback control strategies in biological and engineered systems. We found that two common engineering rules-additive-increase, upon positive feedback, and multiplicative-decrease, upon negative feedback, and multiplicative-increase multiplicative-decrease-are used by diverse biological systems, including for regulating foraging by harvester ant colonies, for maintaining cell-size homeostasis, and for synaptic learning and adaptation in neural circuits. These rules support several goals of these systems, including optimizing efficiency (i.e. using all available resources); splitting resources fairly among cooperating agents, or conversely, acquiring resources quickly among competing agents; and minimizing the latency of responses, especially when conditions change. We hypothesize that theoretical frameworks from distributed computing may offer new ways to analyse adaptation behaviour of biology systems, and in return, biological strategies may inspire new algorithms for discrete-event feedback control in engineering.

What the odor is not: Estimation by elimination

Olfactory systems use a small number of broadly sensitive receptors to combinatorially encode a vast number of odors. We propose a method of decoding such distributed representations by exploiting a statistical fact: Receptors that do not respond to an odor carry more information than receptors that do because they signal the absence of all odorants that bind to them. Thus, it is easier to identify what the odor is not rather than what the odor is. For realistic numbers of receptors, response functions, and odor complexity, this method of elimination turns an underconstrained decoding problem into a solvable one, allowing accurate determination of odorants in a mixture and their concentrations. We construct a neural network realization of our algorithm based on the structure of the olfactory pathway.

Learning probabilistic neural representations with randomly connected circuits

We present a theory of neural circuits’ design and function, inspired by the random connectivity of real neural circuits and the mathematical power of random projections. Specifically, we introduce a family of statistical models for large neural population codes, a straightforward neural circuit architecture that would implement these models, and a biologically plausible learning rule for such circuits. The resulting neural architecture suggests a design principle for neural circuit—namely, that they learn to compute the mathematical surprise of their inputs, given past inputs, without an explicit teaching signal. We applied these models to recordings from large neural populations in monkeys’ visual and prefrontal cortices and show them to be highly accurate, efficient, and scalable.

The brain represents and reasons probabilistically about complex stimuli and motor actions using a noisy, spike-based neural code. A key building block for such neural computations, as well as the basis for supervised and unsupervised learning, is the ability to estimate the surprise or likelihood of incoming high-dimensional neural activity patterns. Despite progress in statistical modeling of neural responses and deep learning, current approaches either do not scale to large neural populations or cannot be implemented using biologically realistic mechanisms. Inspired by the sparse and random connectivity of real neuronal circuits, we present a model for neural codes that accurately estimates the likelihood of individual spiking patterns and has a straightforward, scalable, efficient, learnable, and realistic neural implementation. This model’s performance on simultaneously recorded spiking activity of >100 neurons in the monkey visual and prefrontal cortices is comparable with or better than that of state-of-the-art models. Importantly, the model can be learned using a small number of samples and using a local learning rule that utilizes noise intrinsic to neural circuits. Slower, structural changes in random connectivity, consistent with rewiring and pruning processes, further improve the efficiency and sparseness of the resulting neural representations. Our results merge insights from neuroanatomy, machine learning, and theoretical neuroscience to suggest random sparse connectivity as a key design principle for neuronal computation.

Habituation as a neural algorithm for online odor discrimination

Habituation is a form of simple memory that suppresses neural activity in response to repeated, neutral stimuli. This process is critical in helping organisms guide attention toward the most salient and novel features in the environment. Here, we follow known circuit mechanisms in the fruit fly olfactory system to derive a simple algorithm for habituation. We show, both empirically and analytically, that this algorithm is able to filter out redundant information, enhance discrimination between odors that share a similar background, and improve detection of novel components in odor mixtures. Overall, we propose an algorithmic perspective on the biological mechanism of habituation and use this perspective to understand how sensory physiology can affect odor perception. Our framework may also help toward understanding the effects of habituation in other more sophisticated neural systems.

Neural Circuit Dynamics for Sensory Detection

We consider the question of how sensory networks enable the detection of sensory stimuli in a combinatorial coding space. We are specifically interested in the olfactory system, wherein recent experimental studies have reported the existence of rich, enigmatic response patterns associated with stimulus onset and offset. This study aims to identify the functional relevance of such response patterns (i.e., what benefits does such neural activity provide in the context of detecting stimuli in a natural environment). We study this problem through the lens of normative, optimization-based modeling. Here, we define the notion of a low-dimensional latent representation of stimulus identity, which is generated through action of the sensory network. The objective of our optimization framework is to ensure high-fidelity tracking of a nominal representation in this latent space in an energy-efficient manner. It turns out that the optimal motifs emerging from this framework possess morphologic similarity with prototypical onset and offset responses observed in vivo in locusts (Schistocerca americana) of either sex. Furthermore, this objective can be exactly achieved by a network with reciprocal excitatory-inhibitory competitive dynamics, similar to interactions between projection neurons and local neurons in the early olfactory system of insects. The derived model also makes several predictions regarding maintenance of robust latent representations in the presence of confounding background information and trade-offs between the energy of sensory activity and resultant behavioral measures such as speed and accuracy of stimulus detection.SIGNIFICANCE STATEMENT A key area of study in olfactory coding involves understanding the transformation from high-dimensional sensory stimulus to low-dimensional decoded representation. Here, we examine not only the dimensionality reduction of this mapping but also its temporal dynamics, with specific focus on stimuli that are temporally continuous. Through optimization-based synthesis, we examine how sensory networks can track representations without prior assumption of discrete trial structure. We show that such tracking can be achieved by canonical network architectures and dynamics, and that the resulting responses resemble observations from neurons in the insect olfactory system. Thus, our results provide hypotheses regarding the functional role of olfactory circuit activity at both single neuronal and population scales.

The Drosophila Mushroom Body: From Architecture to Algorithm in a Learning Circuit

The Drosophila brain contains a relatively simple circuit for forming Pavlovian associations, yet it achieves many operations common across memory systems. Recent advances have established a clear framework for Drosophila learning and revealed the following key operations: a) pattern separation, whereby dense combinatorial representations of odors are preprocessed to generate highly specific, nonoverlapping odor patterns used for learning; b) convergence, in which sensory information is funneled to a small set of output neurons that guide behavioral actions; c) plasticity, where changing the mapping of sensory input to behavioral output requires a strong reinforcement signal, which is also modulated by internal state and environmental context; and d) modularization, in which a memory consists of multiple parallel traces, which are distinct in stability and flexibility and exist in anatomically well-defined modules within the network. Cross-module interactions allow for higher-order effects where past experience influences future learning. Many of these operations have parallels with processes of memory formation and action selection in more complex brains.

Biologically plausible deep learning - But how far can we go with shallow networks?

Training deep neural networks with the error backpropagation algorithm is considered implausible from a biological perspective. Numerous recent publications suggest elaborate models for biologically plausible variants of deep learning, typically defining success as reaching around 98% test accuracy on the MNIST data set. Here, we investigate how far we can go on digit (MNIST) and object (CIFAR10) classification with biologically plausible, local learning rules in a network with one hidden layer and a single readout layer. The hidden layer weights are either fixed (random or random Gabor filters) or trained with unsupervised methods (Principal/Independent Component Analysis or Sparse Coding) that can be implemented by local learning rules. The readout layer is trained with a supervised, local learning rule. We first implement these models with rate neurons. This comparison reveals, first, that unsupervised learning does not lead to better performance than fixed random projections or Gabor filters for large hidden layers. Second, networks with localized receptive fields perform significantly better than networks with all-to-all connectivity and can reach backpropagation performance on MNIST. We then implement two of the networks - fixed, localized, random & random Gabor filters in the hidden layer - with spiking leaky integrate-and-fire neurons and spike timing dependent plasticity to train the readout layer. These spiking models achieve >98.2% test accuracy on MNIST, which is close to the performance of rate networks with one hidden layer trained with backpropagation. The performance of our shallow network models is comparable to most current biologically plausible models of deep learning. Furthermore, our results with a shallow spiking network provide an important reference and suggest the use of data sets other than MNIST for testing the performance of future models of biologically plausible deep learning.