Spyridon Chavlis

Spyridon Chavlis, Panayiota Poirazi

Artificial neural networks (ANNs) are at the core of most Deep learning (DL) algorithms that successfully tackle complex problems like image recognition, autonomous driving, and natural language processing. However, unlike biological brains who tackle similar problems in a very efficient manner, DL algorithms require a large number of trainable parameters, making them energy-intensive and prone to overfitting. Here, we show that a new ANN architecture that incorporates the structured connectivity and restricted sampling properties of biological dendrites counteracts these limitations. We find that dendritic ANNs are more robust to overfitting and outperform traditional ANNs on several image classification tasks while using significantly fewer trainable parameters. These advantages are likely the result of a different learning strategy, whereby most of the nodes in dendritic ANNs respond to multiple classes, unlike classical ANNs that strive for class-specificity. Our findings suggest that the incorporation of dendritic properties can make learning in ANNs more precise, resilient, and parameter-efficient and shed new light on how biological features can impact the learning strategies of ANNs.

Kosmas Pinitas, Spyridon Chavlis, Panayiota Poirazi

Current deep learning architectures show remarkable performance when trained in large-scale, controlled datasets. However, the predictive ability of these architectures significantly decreases when learning new classes incrementally. This is due to their inclination to forget the knowledge acquired from previously seen data, a phenomenon termed catastrophic-forgetting. On the other hand, Self-Organizing Maps (SOMs) can model the input space utilizing constrained k-means and thus maintain past knowledge. Here, we propose a novel algorithm inspired by biological neurons, termed Dendritic-Self-Organizing Map (DendSOM). DendSOM consists of a single layer of SOMs, which extract patterns from specific regions of the input space accompanied by a set of hit matrices, one per SOM, which estimate the association between units and labels. The best-matching unit of an input pattern is selected using the maximum cosine similarity rule, while the point-wise mutual information is employed for class inference. DendSOM performs unsupervised feature extraction as it does not use labels for targeted updating of the weights. It outperforms classical SOMs and several state-of-the-art continual learning algorithms on benchmark datasets, such as the Split-MNIST and Split-CIFAR-10. We propose that the incorporation of neuronal properties in SOMs may help remedy catastrophic forgetting.

Eirini Troullinou, Grigorios Tsagkatakis, Spyridon Chavlis et al.

Identification of different neuronal cell types is critical for understanding their contribution to brain functions. Yet, automated and reliable classification of neurons remains a challenge, primarily because of their biological complexity. Typical approaches include laborious and expensive immunohistochemical analysis while feature extraction algorithms based on cellular characteristics have recently been proposed. The former rely on molecular markers, which are often expressed in many cell types, while the latter suffer from similar issues: finding features that are distinctive for each class has proven to be equally challenging. Moreover, both approaches are time consuming and demand a lot of human intervention. In this work we establish the first, automated cell-type classification method that relies on neuronal activity rather than molecular or cellular features. We test our method on a real-world dataset comprising of raw calcium activity signals for four neuronal types. We compare the performance of three different deep learning models and demonstrate that our method can achieve automated classification of neuronal cell types with unprecedented accuracy.