Interpretable by Design: Learning Predictors by Composing Interpretable Queries.

There is a growing concern about typically opaque decision-making with high-performance machine learning algorithms. Providing an explanation of the reasoning process in domain-specific terms can be crucial for adoption in risk-sensitive domains such as healthcare. We argue that machine learning algorithms should be interpretable by design and that the language in which these interpretations are expressed should be domain- and task-dependent. Consequently, we base our model's prediction on a family of user-defined and task-specific binary functions of the data, each having a clear interpretation to the end-user. We then minimize the expected number of queries needed for accurate prediction on any given input. As the solution is generally intractable, following prior work, we choose the queries sequentially based on information gain. However, in contrast to previous work, we need not assume the queries are conditionally independent. Instead, we leverage a stochastic generative model (VAE) and an MCMC algorithm (Unadjusted Langevin) to select the most informative query about the input based on previous query-answers. This enables the online determination of a query chain of whatever depth is required to resolve prediction ambiguities. Finally, experiments on vision and NLP tasks demonstrate the efficacy of our approach and its superiority over post-hoc explanations.

Lens Free Holographic Imaging for Urinary Tract Infection Screening.

OBJECTIVE: The diagnosis of urinary tract infection (UTI) currently requires precise specimen collection, handling infectious human waste, controlled urine storage, and timely transportation to modern laboratory equipment for analysis. Here we investigate holographic lens free imaging (LFI) to show its promise for enabling automatic urine analysis at the patient bedside. METHODS: We introduce an LFI system capable of resolving important urine clinical biomarkers such as red blood cells, white blood cells, crystals, and casts in 2 mm thick urine phantoms. RESULTS: This approach is sensitive to the particulate concentrations relevant for detecting several clinical urine abnormalities such as hematuria and pyuria, linearly correlating to ground truth hemacytometer measurements with R 2 = 0.9941 and R 2 = 0.9973, respectively. We show that LFI can estimate E. coli concentrations of 10 3 to 10 5 cells/mL by counting individual cells, and is sensitive to concentrations of 10 5 cells/mL to 10 8 cells/mL by analyzing hologram texture. Further, LFI measurements of blood cell concentrations are relatively insensitive to changes in bacteria concentrations of over seven orders of magnitude. Lastly, LFI reveals clear differences between UTI-positive and UTI-negative urine from human patients. CONCLUSION: LFI is sensitive to clinically-relevant concentrations of bacteria, blood cells, and other sediment in large urine volumes. SIGNIFICANCE: Together, these results show promise for LFI as a tool for urine screening, potentially offering early, point-of-care detection of UTI and other pathological processes.

Adaptive sparse reconstruction for lensless digital holography via PSF estimation and phase retrieval.

In-line lensless digital holography has great potential in multiple applications; however, reconstructing high-quality images from a single recorded hologram is challenging due to the loss of phase information. Typical reconstruction methods are based on solving a regularized inverse problem and work well under suitable image priors, but they are extremely sensitive to mismatches between the forward model and the actual imaging system. This paper aims to improve the robustness of such algorithms by introducing the adaptive sparse reconstruction method, ASR, which learns a properly constrained point spread function (PSF) directly from data, as opposed to solely relying on physics-based approximations of it. ASR jointly performs holographic reconstruction, PSF estimation, and phase retrieval in an unsupervised way by maximizing the sparsity of the reconstructed images. Like traditional methods, ASR uses the image formation model along with a sparsity prior, which, unlike recent deep learning approaches, allows for unsupervised reconstruction with as little as one sample. Experimental results in synthetic and real data show the advantages of ASR over traditional reconstruction methods, especially in cases where the theoretical PSF does not match that of the actual system.

Generative optical modeling of whole blood for detecting platelets in lens-free images.

In this paper, we consider the task of detecting platelets in images of diluted whole blood taken with a lens-free microscope. Despite having several advantages over traditional microscopes, lens-free imaging systems have the significant challenge that the resolution of the system is typically limited by the pixel dimensions of the image sensor. As a result of this limited resolution, detecting platelets is very difficult even by manual inspection of the images due to the fact that platelets occupy just a few pixels of the reconstructed image. To address this challenge, we develop an optical model of diluted whole blood to generate physically realistic simulated holograms suitable for training machine learning models in a supervised manner. We then use this model to train a convolutional neural network (CNN) for platelet detection and validate our approach by developing a novel optical configuration which allows collecting both lens-free and fluorescent microscopy images of the same field of view of diluted whole blood samples with fluorescently labeled platelets.

Structured Low-Rank Matrix Factorization: Global Optimality, Algorithms, and Applications.

Convex formulations of low-rank matrix factorization problems have received considerable attention in machine learning. However, such formulations often require solving for a matrix of the size of the data matrix, making it challenging to apply them to large scale datasets. Moreover, in many applications the data can display structures beyond simply being low-rank, e.g., images and videos present complex spatio-temporal structures that are largely ignored by standard low-rank methods. In this paper we study a matrix factorization technique that is suitable for large datasets and captures additional structure in the factors by using a particular form of regularization that includes well-known regularizers such as total variation and the nuclear norm as particular cases. Although the resulting optimization problem is non-convex, we show that if the size of the factors is large enough, under certain conditions, any local minimizer for the factors yields a global minimizer. A few practical algorithms are also provided to solve the matrix factorization problem, and bounds on the distance from a given approximate solution of the optimization problem to the global optimum are derived. Examples in neural calcium imaging video segmentation and hyperspectral compressed recovery show the advantages of our approach on high-dimensional datasets.

Multiscale mapping of frequency sweep rate in mouse auditory cortex.

Functional organization is a key feature of the neocortex that often guides studies of sensory processing, development, and plasticity. Tonotopy, which arises from the transduction properties of the cochlea, is the most widely studied organizational feature in auditory cortex; however, in order to process complex sounds, cortical regions are likely specialized for higher order features. Here, motivated by the prevalence of frequency modulations in mouse ultrasonic vocalizations and aided by the use of a multiscale imaging approach, we uncover a functional organization across the extent of auditory cortex for the rate of frequency modulated (FM) sweeps. In particular, using two-photon Ca2+ imaging of layer 2/3 neurons, we identify a tone-insensitive region at the border of AI and AAF. This central sweep region behaves fundamentally differently from nearby neurons in AI and AII, responding preferentially to fast FM sweeps but not to tones or bandlimited noise. Together these findings define a second dimension of organization in the mouse auditory cortex for sweep rate complementary to that of tone frequency.

Multiscale optical Ca2+ imaging of tonal organization in mouse auditory cortex.

Spatial patterns of functional organization, resolved by microelectrode mapping, comprise a core principle of sensory cortices. In auditory cortex, however, recent two-photon Ca2+ imaging challenges this precept, as the traditional tonotopic arrangement appears weakly organized at the level of individual neurons. To resolve this fundamental ambiguity about the organization of auditory cortex, we developed multiscale optical Ca2+ imaging of unanesthetized GCaMP transgenic mice. Single-neuron activity monitored by two-photon imaging was precisely registered to large-scale cortical maps provided by transcranial widefield imaging. Neurons in the primary field responded well to tones; neighboring neurons were appreciably cotuned, and preferred frequencies adhered tightly to a tonotopic axis. By contrast, nearby secondary-field neurons exhibited heterogeneous tuning. The multiscale imaging approach also readily localized vocalization regions and neurons. Altogether, these findings cohere electrode and two-photon perspectives, resolve new features of auditory cortex, and offer a promising approach generalizable to any cortical area.

Conduction block induced by high frequency AC stimulation in unmyelinated nerves.

The potential neurophysiological applications of high frequency AC stimulation (HFAC) in blocking conduction has led to a series of experimental and modeling studies analyzing the effect of HFAC conduction block on mixed nerves. However, many of these computational studies have been based on axon models that are perhaps not valid for the nerves under study. The isolated response of unmyelinated nerves to HFAC has also not been previously studied. In this study, 5-50 kHz sinusoidal HFAC stimulation waveforms were used to reversibly block conduction through the unmyelinated nerve fibers of Aplysia. Unlike myelinated nerves, the minimum HFAC amplitude for blocking conduction in these nerves showed a non-monotonic behavior with frequency. The Hodgkin-Huxley model did not accurately predict the experimentally observed trends but modifying the model to incorporate a frequency-dependent membrane capacitance resulted in a significant change in the high frequency response of the model while still preserving the standard characteristics of action potential propagation.