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.

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.

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.

A neural data structure for novelty detection.

Novelty detection is a fundamental biological problem that organisms must solve to determine whether a given stimulus departs from those previously experienced. In computer science, this problem is solved efficiently using a data structure called a Bloom filter. We found that the fruit fly olfactory circuit evolved a variant of a Bloom filter to assess the novelty of odors. Compared with a traditional Bloom filter, the fly adjusts novelty responses based on two additional features: the similarity of an odor to previously experienced odors and the time elapsed since the odor was last experienced. We elaborate and validate a framework to predict novelty responses of fruit flies to given pairs of odors. We also translate insights from the fly circuit to develop a class of distance- and time-sensitive Bloom filters that outperform prior filters when evaluated on several biological and computational datasets. Overall, our work illuminates the algorithmic basis of an important neurobiological problem and offers strategies for novelty detection in computational systems.

Early Classification of Time Series by Simultaneously Optimizing the Accuracy and Earliness.

The problem of early classification of time series appears naturally in contexts where the data, of temporal nature, are collected over time, and early class predictions are interesting or even required. The objective is to classify the incoming sequence as soon as possible, while maintaining suitable levels of accuracy in the predictions. Thus, we can say that the problem of early classification consists of optimizing two objectives simultaneously: accuracy and earliness. In this context, we present a method for early classification based on combining a set of probabilistic classifiers together with a stopping rule (SR). This SR will act as a trigger and will tell us when to output a prediction or when to wait for more data, and its main novelty lies in the fact that it is built by explicitly optimizing a cost function based on accuracy and earliness. We have selected a large set of benchmark data sets and four other state-of-the-art early classification methods, and we have evaluated and compared our framework obtaining superior results in terms of both earliness and accuracy.

A neural algorithm for a fundamental computing problem.

Similarity search-for example, identifying similar images in a database or similar documents on the web-is a fundamental computing problem faced by large-scale information retrieval systems. We discovered that the fruit fly olfactory circuit solves this problem with a variant of a computer science algorithm (called locality-sensitive hashing). The fly circuit assigns similar neural activity patterns to similar odors, so that behaviors learned from one odor can be applied when a similar odor is experienced. The fly algorithm, however, uses three computational strategies that depart from traditional approaches. These strategies can be translated to improve the performance of computational similarity searches. This perspective helps illuminate the logic supporting an important sensory function and provides a conceptually new algorithm for solving a fundamental computational problem.

Classification accuracies of physical activities using smartphone motion sensors.

BACKGROUND: Over the past few years, the world has witnessed an unprecedented growth in smartphone use. With sensors such as accelerometers and gyroscopes on board, smartphones have the potential to enhance our understanding of health behavior, in particular physical activity or the lack thereof. However, reliable and valid activity measurement using only a smartphone in situ has not been realized. OBJECTIVE: To examine the validity of the iPod Touch (Apple, Inc.) and particularly to understand the value of using gyroscopes for classifying types of physical activity, with the goal of creating a measurement and feedback system that easily integrates into individuals' daily living. METHODS: We collected accelerometer and gyroscope data for 16 participants on 13 activities with an iPod Touch, a device that has essentially the same sensors and computing platform as an iPhone. The 13 activities were sitting, walking, jogging, and going upstairs and downstairs at different paces. We extracted time and frequency features, including mean and variance of acceleration and gyroscope on each axis, vector magnitude of acceleration, and fast Fourier transform magnitude for each axis of acceleration. Different classifiers were compared using the Waikato Environment for Knowledge Analysis (WEKA) toolkit, including C4.5 (J48) decision tree, multilayer perception, naive Bayes, logistic, k-nearest neighbor (kNN), and meta-algorithms such as boosting and bagging. The 10-fold cross-validation protocol was used. RESULTS: Overall, the kNN classifier achieved the best accuracies: 52.3%-79.4% for up and down stair walking, 91.7% for jogging, 90.1%-94.1% for walking on a level ground, and 100% for sitting. A 2-second sliding window size with a 1-second overlap worked the best. Adding gyroscope measurements proved to be more beneficial than relying solely on accelerometer readings for all activities (with improvement ranging from 3.1% to 13.4%). CONCLUSIONS: Common categories of physical activity and sedentary behavior (walking, jogging, and sitting) can be recognized with high accuracies using both the accelerometer and gyroscope onboard the iPod touch or iPhone. This suggests the potential of developing just-in-time classification and feedback tools on smartphones.