Programming and Machine Learning

ABHIJEET SACHDEV

Indian Institute of Technology Mandi +91-962*******

Mandi, H.P – 175001 **************@*****.***

OBJECTIVE

A position in the field of computers with special interests in programming, pattern recognition and kernel

methods.

RESEARCH INTERESTS

I am broadly interested in machine learning, computational statistics and computer vision. My current

research is at the intersection of discriminative learning techniques and computational vision. To this end,

I have been trying to explore the approaches to classification, matching and annotation of scene images for

content based image retrieval (CBIR) of scene images. The CBIR of scene images is the problem of searching

a large repository of scene images according to the content so as to retrieve the imag es relevant to a user’s

query. The focus of my work is on the kernel methods based approaches to classification, matching and

annotation of scene images represented as sets of local feature vectors.

EDUCATION

Indian Institute of Technology Mandi

Master of Science, School of Computing & Electrical Engineering Mandi, H.P, India

CGPA: 7.8/10.0 Present

Government Engineering College Bhopal, M.P, India

Bachelor of Engineering, CSE 2007-2011

Aggregate Score: 69.4%

Sunrisers Higher Secondary School Vidisha, M.P, India

State Board 2005-2006

Aggregate Score: 86.4%

St. Mary’s Senior Secondary School Vidisha, M.P India

State Board 2003-2004

Aggregate Score: 84.4%

Projects Undertaken

Class Specific Hierarchal Classification of Hep-2 Cell Images: (February 2014)

In this project we designed a new framework for classification of HEp-2 cell images. First, we analyzed the

visual characteristics of classes to formulate class -specific image features. Then, realizing that the problem

involves a small number of classes, we treat the classification task as hierarchical verification sub-tasks. Thus,

the overall classification problem is posed as a verification of each class, using its class -specific features.

Hyper-parameter tuning of the kernel function: (May 2014)

The selection of hyper-parameters plays an important role to the performance of least support vector

machines. In this project, a novel hyper-parameter selection method for LS-SVMs is presented. The

proposed method does not need any prior knowledge on the analytic property of the generalization

performance measure and can be used to determine multiple hyper-parameters. The feasibility of this

method is examined on benchmark data sets.

Example Specific Density based Kernel: (August 2014)

In this project, we developed an example-specific density based matching kernel (ESDMK) for the

classification of varying length patterns of long duration speech represented as sets of feature vectors. The

proposed kernel is computed between the pair of examples, represented as sets of feature vectors, by

matching the example specific densities computed at every feature vector in those two examples. In this

work, the density of feature vectors of an example in the volume of K-nearest neighbors of a feature vector is

considered as example specific density. The minimum of the two example -specific densities, one for each

example, at a feature vector is considered as matching score. The ESDMK is then computed as the sum of the

matching score computed at every feature vector in a pair of examples.

Spatial: Example Specific Density based Kernel: (January 2015)

This project is the continuation of the above project. In this project w e worked on scene images. While

designing the Example-Specific density based matching kernel (ESDMK) we did not use the sp atial

information present in the scene images into account. In Spatial: Example-Specific density based matching

kernel (SESDMK) we use spatial information also. We demonstrated its power with standard image data

sets namely Vogel-Schiele, MIT-8-Scene and Corel5k the SESDMK exceeds all other existing kernels in terms

of accuracy, though the designed kernel is having a high complexity as compared to other existing kernels.

PROFESSIONAL EXPERIENCE

Tata Consultancy Services Noida, India

Assistant Software Engineer Oct 2011 – July 2013

Project Name: Walgreens, USA Role:

Developer (Java)

ACHIEVEMENTS

Best team member award in Tata Consultancy Services (India).

Second runner up in C programming contest organized by Cognizant (2007).

All India Rank 528 in Graduate Test of Engineering (GATE - 2013).

TECHNICAL SKILLS

Programming Languages: C, C++, and Java.

Tools: Netbeans, Eclipse, Matlab

Operating Systems: Windows

RELEVANT COURSES

Specialized in: Pattern Recognition, Kernel Methods, Linear Algebra, Statistics, and Optimization.

Publications

Accepted:

Example-Specific Density Based Matching Kernels (ESDMK) for Classification of Scene Images Using

Support Vector Machines, accepted in “International Conference on Image Processing, Computer

Vision, & Pattern Recognition (IPCV 2015)”.

Submitted:

Spatial Example-Specific Density Based Matching Kernels (SESDMK) for Speaker identification and

Speech emotion recognition Using Support Vector Machines, submitted in “Interspeech 2015”.

REFERENCES

Dr. C. Chandra Shekhar Dr. Dileep AD

Professor Associate Professor

School of Computing & Electrical Engineering School of Computing & Electrical Engineer ing

IIT Madras IIT Mandi

Chennai, T.N Mandi, H.P

DECLARATION

I hereby declare that the information provided in this document is true to the best of my knowledge and

belief.

Contact this candidate