M. Sakthi Balan

Peptide Computing - Universality and Complexity

This paper considers a computational model using the peptide-antibody interactions. These interactions which are carried out in parallel can be used to solve NP-complete problems. In this paper we show how to use peptide experiments to solve the Hamiltonian Path Problem (HPP) and a particular version of Set Cover problem called Exact Cover by 3-Sets problem. We also prove that this of model of computation is computationally complete.

Serializing the Parallelism in Parallel Communicating Pushdown Automata Systems

We consider parallel communicating pushdown automata systems (PCPA) and define a property called known communication for it. We use this property to prove that the power of a variant of PCPA, called returning centralized parallel communicating pushdown automata (RCPCPA), is equivalent to that of multi-head pushdown automata. The above result presents a new sub-class of returning parallel communicating pushdown automata systems (RPCPA) called simple-RPCPA and we show that it can be written as a finite intersection of multi-head pushdown automata systems.

On the universality of peptide computing

We present a simulation of Turing machines by peptide–antibody interactions. In contrast to an earlier simulation, this new technique simulates the computation steps automatically by the interaction between peptides and antibodies and does not rely on a “look-and-do” approach, in which the Turing machine program would be interpreted by an extraneous computing agent. We determine the resource requirements of the simulation. Towards a precise definition for peptide computing we construct a new theoretical model. We examine how the simulations presented in this paper fits this model. We also give conditions on the peptide computing model so that it can be simulated by a Turing machine.

Properties of Binding-Blocking Automata: A Study

Binding-Blocking Automata is an automaton model that is inspired by peptide computing. This is a finite state automaton together with the facility to postpone the reading of some symbols to a later part of time by blocking of symbols. In this paper, we study some of the properties of binding-blocking automaton using a measure called blocking quotient and show that any Binding-Blocking Automata can accept only languages where the length grows in a linear fashion. This is the first attempt to study the properties of binding-blocking automata and we feel that this will lead us to investigate further into the properties of this automaton model.

Action Recognition from Optical Flow Visualizations

Optical flow is an important computer vision technique used for motion estimation, object tracking and activity recognition. In this paper, we study the effectiveness of the optical flow feature in recognizing simple actions by using only their RGB visualizations as input to a deep neural network. Feeding only the optical flow visualizations, instead of the raw multimedia content, ensures that only a single motion feature is used as a classification criterion. Here, we deal with human action recognition as a multi-class classification problem. In order to categorize an action, we train an AlexNet-like Convolutional Neural Network (CNN) on Farneback optical flow visualization features of the action videos. We have chosen the KTH data set, which contains six types of action videos, namely walking, running, boxing, jogging, hand-clapping and hand-waving. The accuracy obtained on the test set is 84.72%, and it is naturally less than the state of the art since only a single motion feature is used for classification, but it is high enough to show the effectiveness of optical flow visualization as a good distinguishing criterion for action recognition. The AlexNet-like CNN was trained in Caffe on two NVIDIA Quadro K4200 GPU cards, while the Farneback optical flow features were calculated using OpenCV library.

Binding-blocking automata

We propose an automaton model called as binding-blocking automaton (BBA). It is a finite state automaton together with the ability to block some symbols and postpone them for reading by the head at a later time. The idea of blocking some symbols from being read by the head and unblocking when the system requires to read is motivated by peptide computing where some parts of peptide sequences are blocked by attaching an antibody with higher affinity and unblocked at a later point by the removal of the appropriate antibody. We study the variants of such systems, analyse the power of each variants and show various hierarchy results involving them. We also define normal-forms of binding-blocking automata for one of its variants and also show that the acceptance power of this variant of BBA is strictly less than that of multi-head finite automata.

An event-response model inspired by emotional behaviors

In most attempts to define a system with emotions, researchers have tried to incorporate human emotions in them. In this paper, human emotional behavior is studied in a larger perspective of problem solving. The similarity among reflex action, emotional behavior and common sense behavior is observed. Based on these observations an event-response model for artificial system is introduced.

A study on automation in peptide computing

Peptide computing is a novel way of computing that uses the interaction between peptides and antibodies as a computational model. Since several copies of peptides and antibodies can interact at the same time, this computing model is massively parallel and highly non-deterministic in nature. Due to these advantages this computational model helps us to solve some of the very hard combinatorial problems quite efficiently. Peptide computing involves preparation of peptide sequences and antibodies with respect to the given problem and it also requires the interactions between various peptide sequences and antibodies. To carry out these various operations there is a need for automation. In this paper we enumerate automation issues in peptide computing with respect to several models defined in the literature. We study and address some of the issues and propose some solution to overcome the same.

Peptide computing: universality and theoretical model

We present a new simulation of Turing machines by peptide-antibody interactions. In contrast to a simulation presented previously, this new technique simulates the computation steps automatically and does not rely on a “look-and-do” approach, in which the Turing machine program would be interpreted by an extraneous computing agent. We determine the resource requirements of the simulation. Towards a precise definition for peptide computing we construct a new theoretical model. We examine how the simulations presented in this paper fit this model. We prove that a peptide computing model can be simulated by a Turing machine under certain conditions.

String binding-blocking automata

In a similar way to DNA hybridization, antibodies which specifically recognize peptide sequences can be used for calculation [3,4]. In [4] the concept of peptide computing via peptide-antibody interaction is introduced and an algorithm to solve the satisfiability problem is given. In [3], (1) it is proved that peptide computing is computationally complete and (2) a method to solve two well-known NP-complete problems namely Hamiltonian path problem and exact cover by 3-set problem (a variation of set cover problem) using the interactions between peptides and antibodies is given. In our earlier paper [1], we proposed a theoretical model called as bindingblocking automata (BBA) for computing with peptide-antibody interactions. In [1] we define two types of transitions leftmost(l) and locally leftmost(ll) of BBA and prove that the acceptance power of multihead finite automata is sandwiched between the acceptance power of BBA in l and ll transitions. In this work we define a variant of binding-blocking automata called as string binding-blocking automata and analyze the acceptance power of the new model. The model of binding-blocking automaton can be informally said as a finite state automaton (reading a string of symbols at a time) with (1) blocking and unblocking functions and (2) priority relation in reading of symbols. Blocking and unblocking facilitates skipping 1 some symbols at some instant and reading it when it is necessary. In the sequel we state some results from [1,2] (1) for every BBA there exists an equivalent BBA without priority, (2) for every language accepted by BBA with l transition, there exists BBA with ll transitions accepting the same language, (3) for every language accepted by BBA with l transition there is an equivalent multi-head finite automata which accepts the same language and (4) for every language L accepted by a multi-head finite automaton there is a language L′ accepted by BBA such that L can be written in the form h−1(L′) where h is a homomorphism from L to L′. The basic model of the string binding-blocking automaton is very similar to a BBA but for the blocking and unblocking. Some string of symbols (starting form the head’s position) can be blocked from being read by the head. So only those symbols which are not already read and not blocked can be read by the head. The finite control of the automaton is divided into three sets of states namely blocking states, unblocking states and general reading states. A read symbol can not be read gain, but a blocked symbol can be unblocked and read.