Feature extraction and machine learning techniques for identifying historic urban environmental hazards: New methods to locate lost fossil fuel infrastructure in US cities

U.S. cities contain unknown numbers of undocumented “manufactured gas” sites, legacies of an industry that dominated energy production during the late-19th and early-20th centuries. While many of these unidentified sites likely contain significant levels of highly toxic and biologically persistent contamination, locating them remains a significant challenge. We propose a new method to identify manufactured gas production, storage, and distribution infrastructure in bulk by applying feature extraction and machine learning techniques to digitized historic Sanborn fire insurance maps. Our approach, which relies on a two-part neural network to classify candidate map regions, increases the rate of site identification 20-fold compared to unaided visual coding.

Unsupervised Models of Mouse Visual Cortex

Task-optimized deep convolutional neural networks are the most quantitatively accurate models of the primate ventral visual stream. However, such networks are implausible as a model of the mouse visual system because mouse visual cortex has a known shallower hierarchy and the supervised objectives these networks are typically trained with are likely neither ethologically relevant in content nor in quantity. Here we develop shallow network architectures that are more consistent with anatomical and physiological studies of mouse visual cortex than current models. We demonstrate that hierarchically shallow architectures trained using contrastive objective functions applied to visual-acuity-adapted images achieve neural prediction performance that exceed those of the same architectures trained in a supervised manner and result in the most quantitatively accurate models of the mouse visual system. Moreover, these models' neural predictivity significantly surpasses those of supervised, deep architectures that are known to correspond well to the primate ventral visual stream. Finally, we derive a novel measure of inter-animal consistency, and show that the best models closely match this quantity across visual areas. Taken together, our results suggest that contrastive objectives operating on shallow architectures with ethologically-motivated image transformations may be a biologically-plausible computational theory of visual coding in mice.

Geometry of inter-areal interactions in mouse visual cortex

The response of a set of neurons in an area is the result of the sensory input, the interaction of the neurons within the area as well as the long range interactions between areas. We aimed to study the relation between interactions among multiple areas, and if they are fixed or dynamic. The structural connectivity provides a substrate for these interactions, but anatomical connectivity is not known in sufficient detail and it only gives us a static picture. Using the Allen Brain Observatory Visual Coding Neuropixels dataset, which includes simultaneous recordings of spiking activity from up to 6 hierarchically organized mouse cortical visual areas, we estimate the functional connectivity between neurons using a linear model of responses to flashed static grating stimuli. We characterize functional connectivity between populations via interaction subspaces. We find that distinct subspaces of a source area mediate interactions with distinct target areas, supporting the notion that cortical areas use distinct channels to communicate. Most importantly, using a piecewise linear model for activity within each trial, we find that these interactions evolve dynamically over tens of milliseconds following a stimulus presentation. Inter-areal subspaces become more aligned with the intra-areal subspaces during epochs in which a feedforward wave of activity propagates through visual cortical areas. When the short-term dynamics are averaged over, we find that the interaction subspaces are stable over multiple stimulus blocks. These findings have important implications for understanding how information flows through biological neural networks composed of interconnected modules, each of which may have a distinct functional specialization.

Landmark-Centered Coding in Frontal Cortex Visual Responses

Visual landmarks influence spatial cognition, navigation and goal-directed behavior, but their influence on visual coding for action is poorly understood. Here, we tested landmark influence on prefrontal visual responses by recording from 568 neurons in the frontal (FEF) and supplementary (SEF) eye fields of rhesus macaques. The response field (the area of visual space that modulates activity) for each neuron was tested in the presence of a landmark placed at one of four configurations. We then fit the spatially tuned response fields against a spatial coordinate continuum between gaze- and landmark-centered models. When response fields were fit separately for each target-landmark configuration, the best fits shifted (mean 37% / 40%) toward landmark-centered coding in FEF / SEF respectively, confirming a configuration-dependent intermediate coding scheme. Overall, these data show that external landmarks influence prefrontal visual responses, possibly helping to stabilize movement goals in the presence of noisy internal egocentric signals.

Fixational drift is driven by diffusive dynamics in central neural circuitry

During fixation and between saccades, our eyes undergo diffusive random motion called fixational drift [1]. The role of fixational drift in visual coding and inference has been debated in the past few decades, but the mechanisms that underlie this motion remained unknown. In particular, it has been unclear whether fixational drift arises from peripheral sources, or from central sources within the brain. Here we show that fixational drift is correlated with neural activity, and identify its origin in central neural circuitry within the oculomotor system. We analyzed a large data set of ocular motoneuron (OMN) recordings in the rhesus monkey, alongside precise measurements of eye position [2, 3], and found that most of the variance of fixational eye drifts must arise upstream of the OMNs. The diffusive statistics of the motion points to the oculomotor integrator, a memory circuit responsible for holding the eyes still between saccades, as a likely source of the motion. Theoretical modeling, constrained by the parameters of the primate oculomotor system, supports this hypothesis by accounting for the amplitude as well as the statistics of the motion. Thus, we propose that fixational ocular drift provides a direct observation of diffusive dynamics in a neural circuit responsible for storage of continuous parameter memory in persistent neural activity. The identification of a mechanistic origin for fixational drift is likely to advance the understanding of its role in visual processing and inference.

Design and Applications of GLANCE: GLanceable Alarm Notification for a User Centered Experience

The trade-off between awareness and interruption is a crucial aspect in network fault notifiers: Low severity alarms should not distract operators from other primary tasks, however it might be crucial that operators promptly react to critical notifications. A notification system should hence determine when a particular interruption is appropriate and how it should be presented. In this direction, this paper presents a multistep design path beginning from the objective of designing a proof-of-concept for a glanceable alarm notification component for telecommunication network management systems based on a peripheral display approach. In particular the goal was a notifier guided by severity-based strategies and offering the information expressiveness of a one-notification-at-the-time perspective while enriching it with overview capabilities to guarantee (possibly subliminal) long-term local and global content comprehension and prompt reaction only when the interruption from the foreground task is dictated by the fault severity. A first design macro-phase led to the simple yet effective GLANCE (GLanceable Alarm Notification for a User Centered Experience) model, based on a visual coding technique oriented to comprehension and reaction, and a transition strategy oriented to interruptions and reaction. A second design macro-phase studied the application of GLANCE to a personal customizable multichannel notification tool and to a service-oriented fault monitor for digital terrestrial television broadcasting networks.

Application of colour combinations on visual search tasks under vibration environments

Colour is widely utilised as a visual coding system in visual search, but its application under vibration conditions (e.g., in various vehicles) has not been fully explored. This study was designed to examine the effect of colour combinations on performance of visual search tasks conducted in vibration conditions. Forty-eight university students participated in an experiment where they were required to identify target type and location under 24 colour combinations (half in negative polarity and half in positive polarity) and three vibration conditions (static, low, and high). The findings showed that vibration did not significantly affect performance, perceptions, or physiological aspects. Colour combination significantly affected response time, and the participants preferred colour combinations that had the potential to produce better performance. Colour combinations with negative polarity (e.g., yellow on black and white on black) are recommended for presenting search interfaces. These findings are of importance in human–computer interface designs for information display under vibration conditions.

Landmark-Centered Coding in Frontal Cortex Visual Responses

SummaryVisual landmarks influence spatial cognition [1–3], navigation [4,5] and goal-directed behavior [6–8], but their influence on visual coding in sensorimotor systems is poorly understood [6,9–11]. We hypothesized that visual responses in frontal cortex control gaze areas encode potential targets in an intermediate gaze-centered / landmark-centered reference frame that might depend on specific target-landmark configurations rather than a global mechanism. We tested this hypothesis by recording neural activity in the frontal eye fields (FEF) and supplementary eye fields (SEF) while head-unrestrained macaques engaged in a memory-delay gaze task. Visual response fields (the area of visual space where targets modulate activity) were tested for each neuron in the presence of a background landmark placed at one of four oblique configurations relative to the target stimulus. 102 of 312 FEF and 43 of 256 SEF neurons showed spatially tuned response fields in this task. We then fit these data against a mathematical continuum between a gaze-centered model and a landmark-centered model. When we pooled data across the entire dataset for each neuron, our response field fits did not deviate significantly from the gaze-centered model. However, when we fit response fields separately for each target-landmark configuration, the best fits shifted (mean 37% / 40%) toward landmark-centered coding in FEF / SEF respectively. This confirmed an intermediate gaze / landmark-centered mechanism dependent on local (configuration-dependent) interactions. Overall, these data show that external landmarks influence prefrontal visual responses, likely helping to stabilize gaze goals in the presence of variable eye and head orientations.HighlightsPrefrontal visual responses recorded in the presence of visual landmarksResponse fields showed intermediate gaze / landmark-centered organizationThis influence depended on specific target-landmark configurations

VIP interneurons in mouse primary visual cortex selectively enhance responses to weak but specific stimuli

Vasoactive intestinal peptide-expressing (VIP) interneurons in the cortex regulate feedback inhibition of pyramidal neurons through suppression of somatostatin-expressing (SST) interneurons and, reciprocally, SST neurons inhibit VIP neurons. Although VIP neuron activity in the primary visual cortex (V1) of mouse is highly correlated with locomotion, the relevance of locomotion-related VIP neuron activity to visual coding is not known. Here we show that VIP neurons in mouse V1 respond strongly to low contrast front-to-back motion that is congruent with self-motion during locomotion but are suppressed by other directions and contrasts. VIP and SST neurons have complementary contrast tuning. Layer 2/3 contains a substantially larger population of low contrast preferring pyramidal neurons than deeper layers, and layer 2/3 (but not deeper layer) pyramidal neurons show bias for front-to-back motion specifically at low contrast. Network modeling indicates that VIP-SST mutual antagonism regulates the gain of the cortex to achieve sensitivity to specific weak stimuli without compromising network stability.