- TRACE: Algorithmic ACTS for Preventing the Spread of Recurrent Infectious Diseases on Networks
Biswarup Bhattacharya, Han Ching Ou, Arunesh Sinha, Sze-Chuan Suen, Bistra Dilkina & Milind Tambe
24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD), epiDAMIK: Epidemiology meets Data Mining and Knowledge Discovery Workshop, London, UK (2018) [Oral presentation, Selected as a Health Day spotlight paper for the KDD Poster Reception, Invited to The Knowledge Engineering Review (Cambridge University Press) journal special issue]
PDF / Publication (epiDAMIK) / Abstract / BibTeX / Poster / Slides
Federated AI Meeting (ICML/AAMAS/IJCAI), 16th Adaptive Learning Agents (ALA) Workshop, Stockholm, Sweden (2018) [Oral presentation]
Publication (ALA)
An important means of controlling recurrent infectious diseases is through active screening to detect and treat patients. Disease detection on a large network of individuals is a challenging problem, as the health states of individuals are uncertain and the scale of the problem renders traditional dynamic optimization models impractical. Moreover, efficient use of diagnostic and labor resources is a major concern, especially when the disease is prevalent in a resource-constrained region.
In this paper, we propose a novel active screening model (ACTS) and an algorithm to facilitate active screening for recurrent (no permanent immunity) diseases. Our contributions include: (1) A new approach to modeling SEIS type diseases – diseases with a latent stage and no permanent immunity – using a novel belief-state representation, and (2) a community and eigenvalue-based algorithm (TRACE) to generate an online policy to perform multi-round active screening. We discuss in detail the advantages and disadvantages of existing eigenvaluebased, community-based and greedy approaches towards solving the problem and illustrate the need of developing an algorithmic active screening strategy to achieve better performance scalably. We demonstrate the applicability of TRACE by performing extensive experiments on several real-world publicly available datasets, most of which emulate human contact, with a range of settings to demonstrate applicability to a range of diseases. To the best of our knowledge, this is the first work on developing a multiround active screening model and active screening algorithm for diseases with a latent stage and no permanent immunity.
- Restless Bandits visiting Villages: A Preliminary Study on distributing Public Health Services
Biswarup Bhattacharya
ACM SIGCAS Conference on Computing and Sustainable Societies (COMPASS) [previously ACM DEV], Menlo Park and San Jose, CA, USA (2018) [Oral presentation]
PDF / Abstract / Publication / BibTeX / Slides
Distributing public health services is a major challenge. Health workers are responsible for spreading awareness about preventable health problems among the general public by physically visiting the at-risk individuals. However, a limited number of health workers are often responsible for a large population with a variety of health problems. It is therefore essential to design an effective policy to maximize the coverage and spread of health information with limited resources.
In this paper, we propose a novel hierarchical visitation policy design, scalable to regions of various sizes and diversities, which consists of two levels of planning: i) Macro-level planning (region-level) by adapting the p-functional regions problem (PFRP) to our setting, and ii) Micro-level planning (village-level) by formulating a restless multi-armed bandit (RMAB) model and using POMDPs with Whittle Index Policy. We also consider how to address the heterogeneity of health problems across villages to ensure better service delivery and the dynamic nature of public health priorities, which have not been attempted in previous literature, to the best of our knowledge. Our preliminary experiments show promising results which demonstrate the potential of this methodology to be applied for health policy planning.
- AdGAP: Advanced Global Average Pooling
Arna Ghosh, Biswarup Bhattacharya* & Somnath Basu Roy Chowdhury*
32nd AAAI Conference on Artificial Intelligence (AAAI), Student Abstracts, New Orleans, LA, USA (2018)
PDF / Supplement / Abstract / Publication / BibTeX / Poster / Code
Global average pooling (GAP) has been used previously to generate class activation maps. The motivation behind AdGAP comes from the fact that the convolutional filters possess position information of the essential features and hence, combination of the feature maps could help us locate the class instances in an image. Our novel architecture generates promising results and unlike previous methods, the architecture is not sensitive to the size of the input image, thus promising wider application.
- Training Autoencoders in Sparse Domain
Biswarup Bhattacharya, Arna Ghosh* & Somnath Basu Roy Chowdhury*
32nd AAAI Conference on Artificial Intelligence (AAAI), Student Abstracts, New Orleans, LA, USA (2018)
PDF / Supplement / Abstract / Publication / BibTeX / Poster
Autoencoders (AE) are essential in learning representation of large data (like images) for dimensionality reduction. Images are converted to sparse domain using transforms like Fast Fourier Transform (FFT) or Discrete Cosine Transform (DCT) where information that requires encoding is minimal. By optimally selecting the feature-rich frequencies, we are able to learn the latent vectors more robustly. We successfully show enhanced performance of autoencoders in sparse domain for images.
- INFECT: Infection Estimation in Social Networks
Biswarup Bhattacharya, Iordanis Fostiropoulos, Negarsadat Abolhassani, Qing Dong
University of Southern California, Los Angeles, CA, USA (2018)
PDF / Abstract / BibTeX / Poster
Assessing infection spread accurately to prepare effective mitigation strategies and develop treatment schedules is an important problem. Currently, health workers are interested in receiving accountable data in the shortest time possible to minimize the infection spread in the community. However, collecting data about the health status of people by considering relevant health data such as doctor visits takes time and it results in a gap between the infection burst and protection actions. In this paper, we visualized how contact networks may react to the introduction of infections and identified correlations between predicted health states and actual data using social media data. We first developed an SEIRS disease model which is reasonably accurate in performing realistic disease simulation. Using this realistic disease model, we optimized the disease parameters using historical training data from two social media sites Twitter and Flicker. Then we predicted health states of individuals and compared the results with the online activity or actual regional health information. We then evaluated the model on Los Angeles network and attempted to identify the key reasons behind an area being more prone to a disease. We considered influenza and used our model to perform evaluations on the 2018 flu season.
- Individualized Optimal Behavioral Interventions by Predicting Relapse
Nada Aldarrab, Victor Ardulov, Biswarup Bhattacharya, Su Lei, Han Ching Ou, Yilei Zeng
University of Southern California, Los Angeles, CA, USA (2018)
PDF / Abstract / BibTeX
There exists a number of prevailing therapies and behavioral interventions that are used to help individuals recovering from substance abuse. The Global Appraisal of Individual Needs (GAIN) data set provides information over a large set of adolescents and emerging adults presenting diagnostically relevant demographic variables and reported substance abuse, and relapse frequency. The data are recorded over 4 reporting periods separated evenly over the course of a year (i.e. every 90 days). Of particular interest, GAIN also provides details regarding individuals' attempted treatments for substance abuse during the course of the year. Often treatment options provided are a result of availability rather than specific diagnostic decisions. Our aim is to (1) predict whether an individual will relapse given the treatment received and other demographic and feature information, (2) analyze the effectiveness of treatment options across particular demographic variables, to see if there exist correlations between participant backgrounds and the effectiveness of certain behavioral interventions. The eventual goal is to prescribe personalized treatments in an effort to maximize their likelihood of recovery and minimize their likelihood of relapsing.
- Intent-Aware Contextual Recommendation System
Biswarup Bhattacharya, Iftikhar Burhanuddin, Abhilasha Sancheti & Kushal Satya
17th IEEE International Conference on Data Mining (ICDM), 5th International Workshop on Data Science and Big Data Analytics (DSBDA), New Orleans, LA, USA (2017) [Oral presentation]
PDF / arXiv / Abstract / Publication / BibTeX / Slides
Recommender systems take inputs from user history, use an internal ranking algorithm to generate results and possibly optimize this ranking based on feedback. However, often the recommender system is unaware of the actual intent of the user and simply provides recommendations dynamically without properly understanding the thought process of the user. An intelligent recommender system is not only useful for the user but also for businesses which want to learn the tendencies of their users. Finding out tendencies or intents of a user is a difficult problem to solve.
Keeping this in mind, we sought out to create an intelligent system which will keep track of the user’s activity on a webapplication as well as determine the intent of the user in each session. We devised a way to encode the user’s activity through the sessions. Then, we have represented the information seen by the user in a high dimensional format which is reduced to lower dimensions using tensor factorization techniques. The aspect of intent awareness or intent scoring is dealt with at this stage. Finally, combining the user activity data with the contextual information gives the recommendation score. The final recommendations are then ranked using some filtering and collaborative recommendation techniques to show the top-k recommendations to the user. A provision for feedback is also envisioned in the current system which informs the model to update the various weights involved in the recommender system architecture. Our overall model aims to combine both frequencybased and context-based recommendation systems and quantify the intent of a user to provide better recommendations.
We ran experiments on real-world timestamped user activity data, in the setting of recommending reports to the users of a business analytics tool/application and the results are better than the baselines.We also tuned certain aspects of our model to arrive at optimized results.
- Deep Fault Analysis and Subset Selection in Solar Power Grids
Biswarup Bhattacharya & Abhishek Sinha
31st Neural Information Processing Systems (NeurIPS) Conference, Machine Learning for the Developing World Workshop (ML4D), Long Beach, CA, USA (2017)
PDF / arXiv / Abstract / Publication / BibTeX / Poster
Non-availability of reliable and sustainable electric power is a major problem in the developing world. Renewable energy sources like solar are not very lucrative in the current stage due to various uncertainties like weather, storage, land use among others. There also exists various other issues like mis-commitment of power, absence of intelligent fault analysis, congestion, etc. In this paper, we propose a novel deep learning-based system for predicting faults and selecting power generators optimally so as to reduce costs and ensure higher reliability in solar power systems. The results are highly encouraging and they suggest that the approaches proposed in this paper have the potential to be applied successfully in the developing world.
- Intelligent Fault Analysis in Electrical Power Grids
Biswarup Bhattacharya & Abhishek Sinha
29th IEEE International Conference on Tools with Artificial Intelligence (ICTAI), Boston, MA, USA (2017) [Oral presentation]
PDF / arXiv / Abstract / Publication / BibTeX / Slides
Power grids are one of the most important components of infrastructure in today’s world. Every nation is dependent on the security and stability of its own power grid to provide electricity to the households and industries. A malfunction of even a small part of a power grid can cause loss of productivity, revenue and in some cases even life. Thus, it is imperative to design a system which can detect the health of the power grid and take protective measures accordingly even before a serious anomaly takes place. To achieve this objective, we have set out to create an artificially intelligent system which can analyze the grid information at any given time and determine the health of the grid through the usage of sophisticated formal models and novel machine learning techniques like recurrent neural networks. Our system simulates grid conditions including stimuli like faults, generator output fluctuations, load fluctuations using Siemens PSS/E software and this data is trained using various classifiers like SVM, LSTM and subsequently tested. The results are excellent with our methods giving very high accuracy for the data. This model can easily be scaled to handle larger and more complex grid architectures.
- SIMILARnet: Simultaneous Intelligent Localization and Recognition Network
Arna Ghosh*, Biswarup Bhattacharya* & Somnath Basu Roy Chowdhury*
arXiv Preprint (2017)
PDF / arXiv / Abstract / BibTeX / Code
Global Average Pooling (GAP) [4] has been used previously to generate class activation for image classification tasks. The motivation behind SIMILARnet comes from the fact that the convolutional filters possess position information of the essential features and hence, combination of the feature maps could help us locate the class instances in an image. We propose a biologically inspired model that is free of differential connections and doesn’t require separate training thereby reducing computation overhead. Our novel architecture generates promising results and unlike existing methods, the model is not sensitive to the input image size, thus promising wider application. Codes for the experiment and illustrations can be found at:
https://github.com/brcsomnath/Advanced-GAP.
- Intelligent Subset Selection of Power Generators for Economic Dispatch
Biswarup Bhattacharya & Abhishek Sinha
arXiv Preprint (2017)
PDF / arXiv / Abstract / BibTeX
Sustainable and economical generation of electrical power is an essential and mandatory component of infrastructure in today's world. Optimal generation (generator subset selection) of power requires a careful evaluation of various factors like type of source, generation, transmission & storage capacities, congestion among others which makes this a difficult task. We created a grid to simulate various conditions including stimuli like generator supply, weather and load demand using Siemens PSS/E software and this data is trained using deep learning methods and subsequently tested. The results are highly encouraging. As per our knowledge, this is the first paper to propose a working and scalable deep learning model for this problem.
- Location Optimization of ATM Networks
Somnath Basu Roy Chowdhury, Biswarup Bhattacharya & Sumit Agarwal
arXiv Preprint (2017)
PDF / arXiv / Abstract / BibTeX
ATMs enable the public to perform financial transactions. Banks try to strategically position their ATMs in order to maximize transactions and revenue. In this paper, we introduce a model which provides a score to an ATM location, which serves as an indicator of its relative likelihood of transactions. In order to efficiently capture the spatially dynamic features, we utilize two concurrent prediction models: the local model which encodes the spatial variance by considering highly energetic features in a given location, and the global model which enforces the dominant trends in the entire data and serves as a feedback to the local model to prevent overfitting. The major challenge in learning the model parameters is the lack of an objective function. The model is trained using a synthetic objective function using the dominant features returned from the k-means clustering algorithm in the local model. The results obtained from the energetic features using the models are encouraging.
- Handwriting Profiling using Generative Adversarial Networks
Arna Ghosh*, Biswarup Bhattacharya* & Somnath Basu Roy Chowdhury*
31st AAAI Conference on Artificial Intelligence (AAAI), Student Abstracts, San Francisco, CA, USA (2017)
PDF / arXiv / Abstract / Publication / BibTeX / Poster
Handwriting is a skill learned by humans from a very early age. The ability to develop one's own unique handwriting as well as mimic another person's handwriting is a task learned by the brain with practice. This paper deals with this very problem where an intelligent system tries to learn the handwriting of an entity using Generative Adversarial Networks (GANs). We propose a modified architecture of DCGAN (Radford, Metz, and Chintala 2015) to achieve this. We also discuss about applying reinforcement learning techniques to achieve faster learning. Our algorithm hopes to give new insights in this area and its uses include identification of forged documents, signature verification, computer generated art, digitization of documents among others. Our early implementation of the algorithm illustrates a good performance with MNIST datasets.
- SAD-GAN: Synthetic Autonomous Driving using Generative Adversarial Networks
Arna Ghosh*, Biswarup Bhattacharya* & Somnath Basu Roy Chowdhury*
30th Neural Information Processing Systems (NeurIPS) Conference, Deep Learning for Action and Interaction Workshop, Barcelona, Spain (2016)
PDF / arXiv / Abstract / Publication / BibTeX / Code
Autonomous driving is one of the most recent topics of interest which is aimed at replicating human driving behavior keeping in mind the safety issues. We approach the problem of learning synthetic driving using generative neural networks. The main idea is to make a controller trainer network using images plus key press data to mimic human learning. We used the architecture of a stable GAN to make predictions between driving scenes using key presses. We train our model on one video game (Road Rash) and tested the accuracy and compared it by running the model on other maps in Road Rash to determine the extent of learning.