The modern Boreal summer corresponds to Antarctic winter season and vice versa, while their summer and winter are separated by 6 months due to anti-phase bipolar seesaw in the solar insolation. Similarly, Boreal (65oN) and Austral (65oS) summer insolations suggest an anti-phase relationship, implying warmer Arctic corresponding to cooler Antarctic in the geological past. Sea surface temperature (SST) estimates for the 18000 years BP last glacial maximum (LGM) are suggested to be underestimated by recent studies. Artificial neural networks (ANN) are a branch of artificial intelligence, computer systems that have the ability to learn in a way similar to the mammalian brain. ANN-based SST estimates have been found to be far better than standard methods in paleo climate research. The IIF candidate has generated radiolarian ANN with lower errors in estimates than previously used methods. Therefore, it is envisaged that radiolarian ANN will be used to estimate the paleo-SSTs and salinities for the last 150,000 years BP using high resolution data (2-3 k) from sediment cores from the Arctic and Antarctic Oceans with emphasis on comparison between the Holocene, LGM and Emian interglacial during the tenure of the IIF fellowship under supervision of Prof. Malmgren at Goteborg University, Sweden.We expect that thermal changes in the world ocean at 18K BP reported by CLIMAP would be quite different, which would have a wider implications on boundary conditions used in general circulation models currently used. Radiolarian ANN-based paleo-SSTs and salinities are envisaged to reveal more realistic thermohaline changes in the subtropical and sub-polar regions in both hemispheres, at the strategically located gateway of the thermohaline conveyor belt, which transports tropical h eat to the north Atlantic resulting in warmer winters in the northern Europe than the scenario projected by shutting down of the conveyor belt due abrupt climatic changes.