Active Learning for Discrete Latent Variable Models.

Active learning seeks to reduce the amount of data required to fit the parameters of a model, thus forming an important class of techniques in modern machine learning. However, past work on active learning has largely overlooked latent variable models, which play a vital role in neuroscience, psychology, and a variety of other engineering and scientific disciplines. Here we address this gap by proposing a novel framework for maximum-mutual-information input selection for discrete latent variable regression models. We first apply our method to a class of models known as mixtures of linear regressions (MLR). While it is well known that active learning confers no advantage for linear-gaussian regression models, we use Fisher information to show analytically that active learning can nevertheless achieve large gains for mixtures of such models, and we validate this improvement using both simulations and real-world data. We then consider a powerful class of temporally structured latent variable models given by a hidden Markov model (HMM) with generalized linear model (GLM) observations, which has recently been used to identify discrete states from animal decision-making data. We show that our method substantially reduces the amount of data needed to fit GLM-HMMs and outperforms a variety of approximate methods based on variational and amortized inference. Infomax learning for latent variable models thus offers a powerful approach for characterizing temporally structured latent states, with a wide variety of applications in neuroscience and beyond.

Extracting Low-Dimensional Psychological Representations from Convolutional Neural Networks.

Convolutional neural networks (CNNs) are increasingly widely used in psychology and neuroscience to predict how human minds and brains respond to visual images. Typically, CNNs represent these images using thousands of features that are learned through extensive training on image datasets. This raises a question: How many of these features are really needed to model human behavior? Here, we attempt to estimate the number of dimensions in CNN representations that are required to capture human psychological representations in two ways: (1) directly, using human similarity judgments and (2) indirectly, in the context of categorization. In both cases, we find that low-dimensional projections of CNN representations are sufficient to predict human behavior. We show that these low-dimensional representations can be easily interpreted, providing further insight into how people represent visual information. A series of control studies indicate that these findings are not due to the size of the dataset we used and may be due to a high level of redundancy in the features appearing in CNN representations.

Swine-influenza (H1N1 influenza) mimicking a case of high altitude pulmonary oedema (HAPO).

Here we present a case of a 37-year-old soldier of Indian Army, posted in high altitude area of Ladakh region (>4200 m), who after rejoining after a month leave, presented with dry cough, low-grade fever and dyspnoea on exertion while undergoing acclimatisation, and on examination, was detected to have hypoxaemia, fever and bilateral fine crepitations on chest auscultation. He was started on treatment for high altitude pulmonary oedema at a medical aid post, and later referred to General Hospital at Leh. The course of the illness was complicated by worsening hypoxaemia, continuous high-grade fever, leucopenia and hypotension. Focused medical history revealed that he had travelled to the state of Gujarat during his leave, where high incidence of H1N1 influenza was being reported during the ongoing pandemic. Oseltamivir was empirically started, in addition to parenteral antibiotics and he was started on inotropic support. In view of severe hypoxaemia, not responding to non-invasive ventilation, he was intubated and placed on mechanical ventilation. The patient turned out to be H1N1 positive and succumbed to his illness 3 days later.

Unusual case of chylothorax with unilateral limb swelling.

Here, we present an unusual case of 26-month male toddler who presented with swelling of right lower limb with painless hyperpigmented patch over right groin of 18 months duration associated with recent onset respiratory distress. Evaluation revealed right chylothorax and MRI revealed altered signal intensity in bones and muscles of right lower limb. Lymphoscintigraphy revealed absence of lymphatic channels in right lower limb. Skin biopsy from hyperpigmented patch was suggestive of vasoformative lesion favouring lymphangiomatosis. A diagnosis of Gorham's syndrome was made, and our patient was managed with drainage of chylothorax followed by pleurodesis, parenteral nutrition and radiotherapy.

Diagnostic Evaluation of Skin Adnexal Tumors by Fine-Needle Aspiration Cytology.

BACKGROUND: The diagnosis of skin adnexal tumors is usually based on histopathology. However, fine-needle aspiration cytology (FNAC) can be of great help and shows a high degree of correlation to the final diagnosis. It is helpful in many situations where skin lesions are a manifestation of certain systemic syndromes and aids to the formation of precise treatment plans. MATERIALS AND METHODS: This prospective study was carried out to assess the diagnosis of skin adnexal tumors based on FNAC and their comparison with histopathological diagnoses. There were 14 patients with FNAC and histopathological follow-up in whom a final diagnosis of skin adnexal tumor was made. The results of FNAC were compared with histopathological diagnoses for complete correlation, partial correlation (cases where an FNAC diagnosis of skin adnexal tumor was made but a precise diagnosis of the subtype was not possible) or no correlation (where FNAC failed to diagnose a skin adnexal tumor). RESULTS: Among the 14 cases of skin adnexal tumor, there was total correlation between the FNAC diagnosis and final histopathological diagnosis in 8 cases (57.1%) and a partial correlation in 4 cases (28.5%). There was no correlation of the FNAC diagnosis with the histopathological diagnosis in only 2 cases (14.3%). CONCLUSIONS: FNAC is very useful in making a diagnosis of skin adnexal tumors and helps in the management of the patient.

Familial Disseminated Cutaneous Glomuvenous Malformation: Treatment with Polidocanol Sclerotherapy.

Glomuvenous malformations (GVMs) present as asymptomatic multiple pink-to-blue nodules or plaques. Disseminated lesions are rare, representing 10% of all the cases. Familial cases are caused by mutations in the glomulin gene. A young male presented with multiple bluish-to-dusky red-coloured nodules 10-15 in numbers over the trunk, limbs and buttocks since 12 years of age. They ranged in size from 1 to 3 cm, partially to non-compressible and tender on palpation. There was no history of any systemic complaint. His sister and mother had similar lesions but in a limited distribution. Biopsy showed multiple ectatic dilated vascular channels lined by multiple layers of glomus cells consistent with the diagnosis of GVM. The biopsy of the lesions from the mother and sister also showed similar features. Mutation analysis for glomulin gene could not be done because of the unavailability of the facility at our setting. He underwent sclerotherapy with 3% polidocanol every 2 weeks, and there was significant improvement in the lesions after six sessions of sclerotherapy. The patient is under follow-up and there is no recurrence of the lesions over treated sites after 6 months.

Incidental Diagnosis of Filariasis in Superficial Location by FNAC: A Retrospective Study of 10 Years.

BACKGROUND: Filariasis, transmitted by the bite of various species of mosquito, is a common disease of tropical belt of the world. In South East Asia, including India, it is a major public health problem. The parasite is primarily confined to Lymphatic channel or lymph nodes where it can remain viable for more than two decades. The most common presentation of the disease is asymptomatic/ subclinical microfilaremia, Hydrocele, acute adeno-lymphangitis (ADL) and chronic lymphatic disease. However, the disease may rarely present as superficial swellings at unusual sites. Incidental findings of microfilaria on fine needle aspiration cytology in these situations, helps in prompt recognition of the disease. AIM: The aim of this retrospective study was to assess the incidental diagnosis of Filariasis on fine needle aspiration cytology (FNAC), done for swellings in superficial locations over last ten years at our institute. MATERIALS AND METHODS: This retrospective study was conducted over a period of 10 years from April 2003 to March 2013 on the cases where FNAC was carried out on swellings in superficial locations of the body. Twenty cases of filariasis were diagnosed on routine FNAC material obtained from various superficial sites. Their data was retrieved and analyzed. RESULTS: Among the 20 cases diagnosed as filariasis on FNAC, six cases involved lymph nodes, six involved testiculo-scrotal region, three cases of thyroid swelling, soft tissue swelling in three cases and breast lump in two cases. On FNA smears, microfilariae were seen in all 20 cases, eggs in three cases, adult female worm in two cases and adult male worm in one case. Adherence of inflammatory cells and macrophages to microfilariae was noticed in 4 cases. Eosinophilia was present in seven cases. Microfilaremia was present in only three cases. Causative agent was Wuchereria bancrofti in all cases. CONCLUSION: Although not so common, filariasis should be considered as one of the differential diagnosis of a superficial swelling particularly if clear fluid is obtained on FNAC. Careful screening of smears plays a significant role in recognition of the disease even in asymptomatic patients.