Mark C. Fuhs

A Spin Glass Model of Path Integration in Rat Medial Entorhinal Cortex

Electrophysiological recording studies in the dorsocaudal region of medial entorhinal cortex (dMEC) of the rat reveal cells whose spatial firing fields show a remarkably regular hexagonal grid pattern (Fyhn et al., 2004; Hafting et al., 2005). We describe a symmetric, locally connected neural network, or spin glass model, that spontaneously produces a hexagonal grid of activity bumps on a two-dimensional sheet of units. The spatial firing fields of the simulated cells closely resemble those of dMEC cells. A collection of grids with different scales and/or orientations forms a basis set for encoding position. Simulations show that the animal’s location can easily be determined from the population activity pattern. Introducing an asymmetry in the model allows the activity bumps to be shifted in any direction, at a rate proportional to velocity, to achieve path integration. Furthermore, information about the structure of the environment can be superimposed on the spatial position signal by modulation of the bump activity levels without significantly interfering with the hexagonal periodicity of firing fields. Our results support the conjecture of Hafting et al. (2005) that an attractor network in dMEC may be the source of path integration information afferent to hippocampus.

Context learning in the rodent hippocampus

We present a Bayesian statistical theory of context learning in the rodent hippocampus. While context is often defined in an experimental setting in relation to specific background cues or task demands, we advance a single, more general notion of context that suffices for a variety of learning phenomena. Specifically, a context is defined as a statistically stationary distribution of experiences, and context learning is defined as the problem of how to form contexts out of groups of experiences that cluster together in time. The challenge of context learning is solving the model selection problem: How many contexts make up the rodents world? Solving this problem requires balancing two opposing goals: minimize the variability of the distribution of experiences within a context and minimize the likelihood of transitioning between contexts. The theory provides an understanding of why hippocampal place cell remapping sometimes develops gradually over many days of experience and why even consistent landmark differences may need to be relearned after other environmental changes. The theory provides an explanation for progressive performance improvements in serial reversal learning, based on a clear dissociation between the incremental process of context learning and the relatively abrupt context selection process. The impact of partial reinforcement on reversal learning is also addressed. Finally, the theory explains why alternating sequence learning does not consistently result in unique context-dependent sequence representations in hippocampus.

A visually driven hippocampal place cell model

The firing of place cells in the rodent hippocampus is partly under the control of visual landmarks in the environment (O’Keefe and Conway, 1978). Most place cell models incorporating “visual” input assume noise-free bearing and/or distance information from idealized point objects (e.g., Burgess, Recce, and O’Keefe, 1994; Touretzky and Redish, 1996), rather than attempting to extract landmark information from real-world scenes. This leaves open the question of what kind of visual information is necessary for the hippocampal system to maintain place fields that are stable across trials yet sensitive to landmark position.

Neighbour selection and adaptation for rapid speaker-dependent ASR

Speaker dependent (SD) ASR systems have significantly lower word error rates (WER) compared to speaker independent (SI) systems. However, SD systems require sufficient training data from the target speaker, which is impractical to collect in a short time. We present a technique for training SD models using just few minutes of speakers data. We compensate for the lack of adequate speaker-specific data by selecting neighbours from a database of existing speakers who are acoustically close to the target speaker. These neighbours provide ample training data, which is used to adapt the SI model to obtain an initial SD model for the new speaker with significantly lower WER. We evaluate various neighbour selection algorithms on a large-scale medical transcription task and report significant reduction in WER using only 5 mins of speaker-specific data. We conduct a detailed analysis of various factors such as gender and accent in the neighbour selection. Finally, we study neighbour selection and adaptation in the context of discriminative objective functions.

Detecting bandlimited audio in broadcast television shows

For TV and radio shows containing narrowband speech, Speech-to-text (STT) accuracy on the narrowband audio can be improved by using an acoustic model trained on acoustically matched data. To selectively apply it, one must first be able to accurately detect which audio segments are narrowband. The present paper explores two different bandwidth classi??cation approaches: a traditional Gaussian mixture model (GMM) approach and a spline-based classifier that categorizes audio segments based on their power spectra. We focus on shows found in the DARPA GALE Mandarin training and test sets, where the ratio of wideband to narrowband shows is very large. In this setting, the spline-based classifier reduces the number of misclassified wideband segments by up to 95% relative to the GMM-based classi??er for the same number of misclassified narrowband segments.