Most of the crowd abnormal event detection methods rely on complex hand-crafted features to represent the crowd motion and appearance. Convolutional Neural Networks (CNN) have shown to be a powerful instrument with excellent representational capacities, which can leverage the need for hand-crafted features. In this paper, we show that keeping track of the changes in the CNN feature across time can be used to effectively detect local anomalies. Specifically, we propose to measure local abnormality by combining semantic information (inherited from existing CNN models) with low-level optical-flow. One of the advantages of this method is that it can be used without the fine-tuning phase. The proposed method is validated on challenging abnormality detection datasets and the results show the superiority of our approach compared with the state-of-theart methods.

Due to their small dimensions, electrophysiology on thin and intricate axonal branches in support of understanding their role in normal and diseased brain function poses experimental challenges. To reduce experimental complexity, we coupled microelectrode arrays (MEAs) to bi-level microchannel devices for the long-term in vitro tracking of axonal morphology and activity with high spatiotemporal resolution. Our model allowed the long-term multisite recording from pure axonal branches in a microscopy-compatible environment. Compartmentalizing the network structure into interconnected subpopulations simplified access to the locations of interest. Electrophysiological data over 95 days in vitro (DIV) showed an age-dependent increase of axonal conduction velocity, which was positively correlated with, but independent of evolving burst activity over time. Conduction velocity remained constant at chemically increased network activity levels. In contrast, low frequency (1 Hz, 180 repetitions) electrical stimulation of axons or network subpopulations evoked amplitude-dependent direct (5–35 ms peri-stimulus) and polysynaptic (35–1,000 ms peri-stimulus) activity with temporarily (<35 ms) elevated propagation velocities along the perisomatic branches. Furthermore, effective stimulation amplitudes were found to be significantly lower (>250 mV) in microchannels when compared with those reported for unconfined cultures (>800 mV). The experimental paradigm may lead to new insights into stimulation-induced axonal plasticity.

No abstract

No abstract

No abstract

No abstract

We investigate a classification problem using multiple mobile agents capable of collecting (partial) posedependent observations of an unknown environment. The objective is to classify an image over a finite time horizon. We propose a network architecture on how agents should form a local belief, take local actions, and extract relevant features from their raw partial observations. Agents are allowed to exchange information with their neighboring agents to update their own beliefs. It is shown how reinforcement learning techniques can be utilized to achieve decentralized implementation of the classification problem by running a decentralized consensus protocol. Our experimental results on the MNIST handwritten digit dataset demonstrates the effectiveness of our proposed framework.

No abstract