Interpretable Compositional Representations for Robust Few-Shot Generalization.

We propose Recognition as Part Composition (RPC), an image encoding approach inspired by human cognition. It is based on the cognitive theory that humans recognize complex objects by components, and that they build a small compact vocabulary of concepts to represent each instance with. RPC encodes images by first decomposing them into salient parts, and then encoding each part as a mixture of a small number of prototypes, each representing a certain concept. We find that this type of learning inspired by human cognition can overcome hurdles faced by deep convolutional networks in low-shot generalization tasks, like zero-shot learning, few-shot learning and unsupervised domain adaptation. Furthermore, we find a classifier using an RPC image encoder is fairly robust to adversarial attacks, that deep neural networks are known to be prone to. Given that our image encoding principle is based on human cognition, one would expect the encodings to be interpretable by humans, which we find to be the case via crowd-sourcing experiments. Finally, we propose an application of these interpretable encodings in the form of generating synthetic attribute annotations for evaluating zero-shot learning methods on new datasets.

Machine learning combining CT findings and clinical parameters improves prediction of length of stay and ICU admission in torso trauma.

OBJECTIVE: To develop machine learning (ML) models capable of predicting ICU admission and extended length of stay (LOS) after torso (chest, abdomen, or pelvis) trauma, by using clinical and/or imaging data. MATERIALS AND METHODS: This was a retrospective study of 840 adult patients admitted to a level 1 trauma center after injury to the torso over the course of 1 year. Clinical parameters included age, sex, vital signs, clinical scores, and laboratory values. Imaging data consisted of any injury present on CT. The two outcomes of interest were ICU admission and extended LOS, defined as more than the median LOS in the dataset. We developed and tested artificial neural network (ANN) and support vector machine (SVM) models, and predictive performance was evaluated by area under the receiver operating characteristic (ROC) curve (AUC). RESULTS: The AUCs of SVM and ANN models to predict ICU admission were up to 0.87 ± 0.03 and 0.78 ± 0.02, respectively. The AUCs of SVM and ANN models to predict extended LOS were up to 0.80 ± 0.04 and 0.81 ± 0.05, respectively. Predictions based on imaging alone or imaging with clinical parameters were consistently more accurate than those based solely on clinical parameters. CONCLUSIONS: The best performing models incorporated imaging findings and outperformed those with clinical findings alone. ML models have the potential to help predict outcomes in trauma by integrating clinical and imaging findings, although further research may be needed to optimize their performance. KEY POINTS: • Artificial neural network and support vector machine-based models were used to predict the intensive care unit admission and extended length of stay after trauma to the torso. • Our input data consisted of clinical parameters and CT imaging findings derived from radiology reports, and we found that combining the two significantly enhanced the prediction of both outcomes with either model. • The highest accuracy (83%) and highest area under the receiver operating characteristic curve (0.87) were obtained for artificial neural networks and support vector machines, respectively, by combining clinical and imaging features in the prediction of intensive care unit admission.

Unique contributions of parvalbumin and cholinergic interneurons in organizing striatal networks during movement.

Striatal parvalbumin (PV) and cholinergic interneurons (CHIs) are poised to play major roles in behavior by coordinating the networks of medium spiny cells that relay motor output. However, the small numbers and scattered distribution of these cells have hindered direct assessment of their contribution to activity in networks of medium spiny neurons (MSNs) during behavior. Here, we build on recent improvements in single-cell calcium imaging combined with optogenetics to test the capacity of PVs and CHIs to affect MSN activity and behavior in mice engaged in voluntary locomotion. We find that PVs and CHIs have unique effects on MSN activity and dissociable roles in supporting movement. PV cells facilitate movement by refining the activation of MSN networks responsible for movement execution. CHIs, in contrast, synchronize activity within MSN networks to signal the end of a movement bout. These results provide new insights into the striatal network activity that supports movement.

Mild Blast Injury Produces Acute Changes in Basal Intracellular Calcium Levels and Activity Patterns in Mouse Hippocampal Neurons.

Mild traumatic brain injury (mTBI) represents a serious public health concern. Although much is understood about long-term changes in cell signaling and anatomical pathologies associated with mTBI, little is known about acute changes in neuronal function. Using large scale Ca2+ imaging in vivo, we characterized the intracellular Ca2+ dynamics in thousands of individual hippocampal neurons using a repetitive mild blast injury model in which blasts were directed onto the cranium of unanesthetized mice on two consecutive days. Immediately following each blast event, neurons exhibited two types of changes in Ca2+ dynamics at different time scales. One was a reduction in slow Ca2+ dynamics that corresponded to shifts in basal intracellular Ca2+ levels at a time scale of minutes, suggesting a disruption of biochemical signaling. The second was a reduction in the rates of fast transient Ca2+ fluctuations at the sub-second time scale, which are known to be closely linked to neural activity. Interestingly, the blast-induced changes in basal Ca2+ levels were independent of the changes in the rates of fast Ca2+ transients, suggesting that blasts had heterogeneous effects on different cell populations. Both types of changes recovered after ∼1 h. Together, our results demonstrate that mTBI induced acute, heterogeneous changes in neuronal function, altering intracellular Ca2+ dynamics across different time scales, which may contribute to the initiation of longer-term pathologies.

Sequential Optimization for Efficient High-Quality Object Proposal Generation.

We are motivated by the need for a generic object proposal generation algorithm which achieves good balance between object detection recall, proposal localization quality and computational efficiency. We propose a novel object proposal algorithm, BING++, which inherits the virtue of good computational efficiency of BING [1] but significantly improves its proposal localization quality. At high level we formulate the problem of object proposal generation from a novel probabilistic perspective, based on which our BING++ manages to improve the localization quality by employing edges and segments to estimate object boundaries and update the proposals sequentially. We propose learning the parameters efficiently by searching for approximate solutions in a quantized parameter space for complexity reduction. We demonstrate the generalization of BING++ with the same fixed parameters across different object classes and datasets. Empirically our BING++ can run at half speed of BING on CPU, but significantly improve the localization quality by 18.5 and 16.7 percent on both VOC2007 and Microhsoft COCO datasets, respectively. Compared with other state-of-the-art approaches, BING++ can achieve comparable performance, but run significantly faster.

Prediction of hospitalization due to heart diseases by supervised learning methods.

BACKGROUND: In 2008, the United States spent $2.2 trillion for healthcare, which was 15.5% of its GDP. 31% of this expenditure is attributed to hospital care. Evidently, even modest reductions in hospital care costs matter. A 2009 study showed that nearly $30.8 billion in hospital care cost during 2006 was potentially preventable, with heart diseases being responsible for about 31% of that amount. METHODS: Our goal is to accurately and efficiently predict heart-related hospitalizations based on the available patient-specific medical history. To the best of our knowledge, the approaches we introduce are novel for this problem. The prediction of hospitalization is formulated as a supervised classification problem. We use de-identified Electronic Health Record (EHR) data from a large urban hospital in Boston to identify patients with heart diseases. Patients are labeled and randomly partitioned into a training and a test set. We apply five machine learning algorithms, namely Support Vector Machines (SVM), AdaBoost using trees as the weak learner, logistic regression, a naïve Bayes event classifier, and a variation of a Likelihood Ratio Test adapted to the specific problem. Each model is trained on the training set and then tested on the test set. RESULTS: All five models show consistent results, which could, to some extent, indicate the limit of the achievable prediction accuracy. Our results show that with under 30% false alarm rate, the detection rate could be as high as 82%. These accuracy rates translate to a considerable amount of potential savings, if used in practice.

Behavior subtraction.

Background subtraction has been a driving engine for many computer vision and video analytics tasks. Although its many variants exist, they all share the underlying assumption that photometric scene properties are either static or exhibit temporal stationarity. While this works in many applications, the model fails when one is interested in discovering changes in scene dynamics instead of changes in scene's photometric properties; the detection of unusual pedestrian or motor traffic patterns are but two examples. We propose a new model and computational framework that assume the dynamics of a scene, not its photometry, to be stationary, i.e., a dynamic background serves as the reference for the dynamics of an observed scene. Central to our approach is the concept of an event, which we define as short-term scene dynamics captured over a time window at a specific spatial location in the camera field of view. Unlike in our earlier work, we compute events by time-aggregating vector object descriptors that can combine multiple features, such as object size, direction of movement, speed, etc. We characterize events probabilistically, but use low-memory, low-complexity surrogates in a practical implementation. Using these surrogates amounts to behavior subtraction, a new algorithm for effective and efficient temporal anomaly detection and localization. Behavior subtraction is resilient to spurious background motion, such as due to camera jitter, and is content-blind, i.e., it works equally well on humans, cars, animals, and other objects in both uncluttered and highly cluttered scenes. Clearly, treating video as a collection of events rather than colored pixels opens new possibilities for video analytics.

Activity based matching in distributed camera networks.

In this paper, we consider the problem of finding correspondences between distributed cameras that have partially overlapping field of views. When multiple cameras with adaptable orientations and zooms are deployed, as in many wide area surveillance applications, identifying correspondence between different activities becomes a fundamental issue. We propose a correspondence method based upon activity features that, unlike photometric features, have certain geometry independence properties. The proposed method is robust to pose, illumination and geometric effects, unsupervised (does not require any calibration objects). In addition, these features are amenable to low communication bandwidth and distributed network applications. We present quantitative and qualitative results with synthetic and real life examples, and compare the proposed method with scale invariant feature transform (SIFT) based method. We show that our method significantly outperforms the SIFT method when cameras have significantly different orientations. We then describe extensions of our method in a number of directions including topology reconstruction, camera calibration, and distributed anomaly detection.