We extend the multinomial logit model to represent some of the empirical phenomena that are frequently observed in the choices made by humans. These phenomena include the similarity effect, the attraction effect, and the compromise effect. We formally quantify the strength of these phenomena that can be represented by our choice model, which illuminates the flexibility of our choice model. We then show that our choice model can be represented as a restricted Boltzmann machine and that its parameters can be learned effectively from data. Our numerical experiments with real data of human choices suggest that we can train our choice model in such a way that it represents the typical phenomena of choice.

We introduce a novel probabilistic tracking algorithm that incorporates combinatorial data association constraints and model-based track management using variational Bayes. We use a Bethe entropy approximation to incorporate data association constraints that are often ignored in previous probabilistic tracking algorithms. Noteworthy aspects of our method include a model-based mechanism to replace heuristic logic typically used to initiate and destroy tracks, and an assignment posterior with linear computation cost in window length as opposed to the exponential scaling of previous MAP-based approaches. We demonstrate the applicability of our method on radar tracking and computer vision problems. The field of tracking is broad and possesses many applications, particularly in radar/sonar [1], robotics [14], and computer vision [3].

Price posting is a common selling mechanism across various corners of e-commerce.

A core challenge is to generalize the manipulation skills to objects in different locations. We hypothesize that modeling the relational information in the environment can significantly improve generalization.

W e restate some of the definitions from the paper for convenience. Moreover, depending on the algorithm implementation, the number of iterations can be finite (e.g. by After the allowable time steps ended, the simulation was reset to its initial state. The main goal is to tag as many opponents as possible within a given time frame. The states reflect the agent's current position and the opponents' positions. The Baby POMDP is a classic problem that represents the scenario of a baby and a caregiver. The states in this problem represent the baby's needs, which could be hunger, The observations are binary, either the baby is crying or not.

Autonomous agents operating in real-world scenarios frequently encounter uncertainty and make decisions based on incomplete information.

Despite the success of neural-based combinatorial optimization methods for end-to-end heuristic learning, out-of-distribution generalization remains a challenge.