<p>The current research on the Internet of Things (IoT) emphasizes integrating cognition into system design and architecture, leading to the development of a new discipline called cognitive IoT (CIoT). CIoT takes on many of the traits and challenges of the Internet of Things. In addition, the CIoT has many sensors. These&#xa0;sensors detect and share information for various purposes, requiring the fusion of information intelligently. When sensors detect, decide, and react to perception, the cognitively-inspired technique is necessary for multisensory data fusion, which must be computationally efficient. Therefore, this research offers a method for combining cognitive-inspired fusion approaches with the probabilistic analysis of sensory input. The proposed method calls for passing the sensory data via the total variation regularizer to bring the noise level down to a tolerable level before applying probabilistic clustering to fix the problem of missing entries. The next step in determining the plausibility value is to calculate the probability value for each of the generated clusters. When deciding which cluster to fuse (used for intelligent set formation), the one with the highest plausibility value is taken into consideration. Further, the combination of fused data is used for the entropy computation. Thus, the most plausible patterns are retrieved from interesting sets by picking the row with the highest entropy. Experimental evaluations utilizing 21.25&#xa0;years of weather data show that the proposed method outperforms competing alternatives (accuracy is greater than 99%), as confirmed by cross-validation using several scales.</p>