 Hi everyone, I am Taposh. This is a short presentation of our work Compact EFI for unbounded attribute weighted sums for log space from SSDH. It's a joint work with Pothich Dato and Katshawaki Takashima. In function encryption, the setup authority generates a master public key MPK and a master secret key MSK. Using MSK, the authority computes the secret key SKF corresponding to any function F and distributes the secret key to the user of the system. Using MPK, a data owner encrypts his or her data and publishes the ciphertext CTM corresponding to the message M. Finally, a user decrypts CTM using the secret key SKF and learns the function of the message M F of M. Simulation security is the most desired security model for FE, which roughly says that, given functional secret keys SKF1, SKF2, SKF3, the adversary can only recover the functional values of the message M and nothing else about the message from the ciphertext CTM. FE has been built for general class of functions, such as student machines or all circuits. However, existing constructions are inefficient and are impractical. These are bounded collision resistance and depends on heavy cryptographic tools such as indistinguishable obfuscation or sub-exponential security of LW. As a result, the community also focused on building FE for specific class of functions. For example, linear quadratic functions and their variants. These constructions are efficient and fully collision resistant. The security is based on standard assumptions like DDH, KLEAN or LW. These schemes can also be deployed in practice. This work focuses on building FE for a specific class of functions, namely attribute weighted sums or AWS, introduced by Abdullah Gong and we. Here, secret keys are generated for functions which take input a vector of length M and outputs a vector of length N. The message is a tackle of X and Z, where X is public and X, Z and Z is private. The functionality outputs the inner product between the vectors F of X and Z. Prior works build the primitive for function class AVP or arithmetic branching program. The master publicly depends on the lengths of X and Z. The cypher tech size depends on the length of Z. Although the first work of Abdullah et al. considered only semi-adaptive security, last year we achieved adaptive simulation security. The functionality also has several applications. For example, it captures existing FE schemes such as IPFE, ABE, ABIPFE. However, the current structure of AWS has many limitations. It only considers non-uniform or non-dynamic model. The setup is bounded. The cypher tech size is not input specific and it captures only bounded FE schemes. Let's see how to remove all the limitations one by one. Firstly, we can consider uniform model of computation such as student machines. Setup could be made unbounded. That is, MPK only depends on security parameter. The cypher tech size is input specific. We also hope to have adaptive simulation security. This is exactly the motivation of our work. Due to all these positive features, the functionality also captures unbounded FE schemes such as unbounded IPFE or unbounded ABE. In summary, we define FE for unbounded AWS and construct it for lock space during machines. Our scheme enjoys exciting properties like dynamism meaning that attributes can be added or deleted at any point of time. The cypher text is compact meaning that it does not grow with the multiple occurrence of a particular attribute in the weight function. Our scheme is fully collision resistant and adaptively simulation secure under the well-studded A6DS assumption. Thanks for watching. Please attend my full talk at the conference. You can also see our paper at a print.