Neural mechanisms of credit assignment for inferred relationships in a structured world

Animals have been proposed to abstract compact representations of a task's structure that could, in principle, support accelerated learning and flexible behavior. Whether and how such abstracted representations may be used to assign credit for inferred, but unobserved, relationships in structured environments are unknown. Here, we develop a novel hierarchical reversal-learning task and Bayesian learning model to assess the computational and neural mechanisms underlying how humans infer specific choice-outcome associations via structured knowledge. We find that the medial prefrontal cortex (mPFC) efficiently represents hierarchically related choice-outcome associations governed by the same latent cause, using a generalized code to assign credit for both experienced and inferred outcomes. Furthermore, mPFC and lateral orbital frontal cortex track the inferred current "position" within a latent association space that generalizes over stimuli. Collectively, these findings demonstrate the importance both of tracking the current position in an abstracted task space and efficient, generalizable representations in prefrontal cortex for supporting flexible learning and inference in structured environments.

Two Extensions of Cover Automata

Deterministic Finite Cover Automata (DFCA) are compact representations of finite languages. Deterministic Finite Automata with “do not care” symbols and Multiple Entry Deterministic Finite Automata are both compact representations of regular languages. This paper studies the benefits of combining these representations to get even more compact representations of finite languages. DFCAs are extended by accepting either “do not care” symbols or considering multiple entry DFCAs. We study for each of the two models the existence of the minimization or simplification algorithms and their computational complexity, the state complexity of these representations compared with other representations of the same language, and the bounds for state complexity in case we perform a representation transformation. Minimization for both models proves to be NP-hard. A method is presented to transform minimization algorithms for deterministic automata into simplification algorithms applicable to these extended models. DFCAs with “do not care” symbols prove to have comparable state complexity as Nondeterministic Finite Cover Automata. Furthermore, for multiple entry DFCAs, we can have a tight estimate of the state complexity of the transformation into equivalent DFCA.

Fast and compact matching statistics analytics

Fast, lightweight methods for comparing the sequence of ever larger assembled genomes from ever growing databases are increasingly needed in the era of accurate long reads and pan-genome initiatives. Matching statistics is a popular method for computing whole-genome phylogenies and for detecting structural rearrangements between two genomes, since it is amenable to fast implementations that require a minimal setup of data structures. However, current implementations use a single core, take too much memory to represent the result, and do not provide efficient ways to analyze the output in order to explore local similarities between the sequences. We develop practical tools for computing matching statistics between large-scale strings, and for analyzing its values, faster and using less memory than the state of the art. Specifically, we design a parallel algorithm for shared-memory machines that computes matching statistics 30 times faster with 48 cores in the cases that are most difficult to parallelize. We design a lossy compression scheme that shrinks the matching statistics array to a bitvector that takes from 0.8 to 0.2 bits per character, depending on the dataset and on the value of a threshold, and that achieves 0.04 bits per character in some variants. And we provide efficient implementations of range-maximum and range-sum queries that take a few tens of milliseconds while operating on our compact representations, and that allow computing key local statistics about the similarity between two strings. Our toolkit makes construction, storage, and analysis of matching statistics arrays practical for multiple pairs of the largest genomes available today, possibly enabling new applications in comparative genomics.

The most robust representations of flow trajectories are Lagrangian coherent structures

What is the most robust way to communicate flow trajectories? To answer this question, we employ two neural networks to respectively deconstruct (the encoder) and reconstruct (the decoder) trajectories, where information is passed between the two networks through a low-dimensional latent space in a set-up known as an autoencoder. To ensure that their communications are robust, we add noise to the coded information passed through this latent space. In the low-noise limit the latent space structures are non-spatial in nature, resembling modes of a principle component analysis (PCA). However, as the signal-to-noise ratio is decreased, we uncover Lagrangian coherent structures (LCS) as the most compact representations which still allow the decoder to accurately reconstruct trajectories. This relationship offers increased interpretability to both PCA and LCS analysis, and helps to bridge the gap between two methods of flow analysis.

Mode-assisted joint training of deep Boltzmann machines

The deep extension of the restricted Boltzmann machine (RBM), known as the deep Boltzmann machine (DBM), is an expressive family of machine learning models which can serve as compact representations of complex probability distributions. However, jointly training DBMs in the unsupervised setting has proven to be a formidable task. A recent technique we have proposed, called mode-assisted training, has shown great success in improving the unsupervised training of RBMs. Here, we show that the performance gains of the mode-assisted training are even more dramatic for DBMs. In fact, DBMs jointly trained with the mode-assisted algorithm can represent the same data set with orders of magnitude lower number of total parameters compared to state-of-the-art training procedures and even with respect to RBMs, provided a fan-in network topology is also introduced. This substantial saving in number of parameters makes this training method very appealing also for hardware implementations.

Spiking Autoencoders With Temporal Coding

Spiking neural networks with temporal coding schemes process information based on the relative timing of neuronal spikes. In supervised learning tasks, temporal coding allows learning through backpropagation with exact derivatives, and achieves accuracies on par with conventional artificial neural networks. Here we introduce spiking autoencoders with temporal coding and pulses, trained using backpropagation to store and reconstruct images with high fidelity from compact representations. We show that spiking autoencoders with a single layer are able to effectively represent and reconstruct images from the neuromorphically-encoded MNIST and FMNIST datasets. We explore the effect of different spike time target latencies, data noise levels and embedding sizes, as well as the classification performance from the embeddings. The spiking autoencoders achieve results similar to or better than conventional non-spiking autoencoders. We find that inhibition is essential in the functioning of the spiking autoencoders, particularly when the input needs to be memorised for a longer time before the expected output spike times. To reconstruct images with a high target latency, the network learns to accumulate negative evidence and to use the pulses as excitatory triggers for producing the output spikes at the required times. Our results highlight the potential of spiking autoencoders as building blocks for more complex biologically-inspired architectures. We also provide open-source code for the model.

Automatic Narrative Summarization for Visualizing Cyber Security Logs and Incident Reports

Cyber security logs and incident reports describe a narrative, but in practice analysts view the data in tables where it can be difficult to follow the narrative. Narrative visualizations are useful, but common examples use a summarized narrative instead of the full story's narrative; it is unclear how to automatically generate these summaries. This paper presents (1) a narrative summarization algorithm to reduce the size and complexity of cyber security narratives with a user-customizable summarization level, and (2) a narrative visualization tailored for incident reports and network logs. An evaluation on real incident reports shows that the summarization algorithm reduces false positives and improves average precision by 41% while reducing average incident report size up to 79%. Together, the visualization and summarization algorithm generate compact representations of cyber narratives that earned praise from a SOC analyst. We further demonstrate that the summarization algorithm can apply to other types of dynamic graphs by automatically generating a summary of the Les Misérables character interaction graph. We find that the list of main characters in the automatically generated summary has substantial agreement with human-generated summaries. A version of this paper, data, and code is freely available at https://osf.io/ekzbp/.

Assessing the combination of DistilBERT news representations and difusion topological features to classify fake news

Fake news (FN) have affected people’s lives in unimaginable ways. The automatic classification of FN is a vital tool to prevent their dissemination and support fact-checking. Related work has shown that FN spread faster, deeper, and more broadly than truthful news on social media. Deep learning has produced state-of-the-art solutions in this field, mainly based on textual attributes. In this paper, we propose to combine compact representations of the textual news properties generated using DistilBERT, with topological metrics extracted from their propagation network in social media. Using a dataset related to politics and distinct learning algorithms, we extensively assessed the components of the proposed solution. Regarding the textual attributes, we reached results comparable to stateof-the-art solutions using only the news title and contents, which is useful for FN early detection. We assessed the influential topological metrics, and the effect of their combination with the news textual features. We also explored the use of ensembles. Our results were very promising, revealing the potential of the features proposed and the adoption of ensembles.

Neural-Network-Based Cross-Channel Intra Prediction

To reduce the redundancy among different color channels, e.g., YUV, previous methods usually adopt a linear model that tends to be oversimple for complex image content. We propose a neural-network-based method for cross-channel prediction in intra frame coding. The proposed network utilizes twofold cues, i.e., the neighboring reconstructed samples with all channels, and the co-located reconstructed samples with partial channels. Specifically, for YUV video coding, the neighboring samples with YUV are processed by several fully connected layers; the co-located samples with Y are processed by convolutional layers; and the proposed network fuses the twofold cues. We observe that the integration of twofold information is crucial to the performance of intra prediction of the chroma components. We have designed the network architecture to achieve a good balance between compression performance and computational efficiency. Moreover, we propose a transform domain loss for the training of the network. The transform domain loss helps obtain more compact representations of residues in the transform domain, leading to higher compression efficiency. The proposed method is plugged into HEVC and VVC test models to evaluate its effectiveness. Experimental results show that our method provides more accurate cross-channel intra prediction compared with previous methods. On top of HEVC, our method achieves on average 1.3%, 5.4%, and 3.8% BD-rate reductions for Y, Cb, and Cr on common test sequences, and on average 3.8%, 11.3%, and 9.0% BD-rate reductions for Y, Cb, and Cr on ultra-high-definition test sequences. On top of VVC, our method achieves on average 0.5%, 1.7%, and 1.3% BD-rate reductions for Y, Cb, and Cr on common test sequences.