In this paper, we consider an M/G/1 retrial queue with Bernoulli feedback and general retrial times where the external arrivals blocked either with probability q join the infinite waiting room called non-retrial queue or with probability p, (p+q = 1) leave the service area and enter the retrial group called orbit in FIFO discipline. Also we assume that only the customer at the head of the orbit...

あらまし 深層学習の重要な一要素として,レイヤーワイズの pre-trainingがある.レイヤーワイズの pre-trainingの 一つとして制約つきBoltzmannマシン (RBM)が有名である.RBMにはBernoulli-Bernoulli型とGaussian-Bernoulli 型があり,従来法の学習アルゴリズムとしてContrastive Divergence法が有名である.本発表ではBernoulli-Bernoulli 型,Gaussian-Bernoulli型の両方の最尤学習アルゴリズムと,最適化規準として新たに最大再構築確率を導入し,そ の学習アルゴリズムに焦点を当て,経験的に高速で安定に収束する補助関数法による新たな更新アルゴリズムの導出 を行う.そして,人工データによる収束性能の比較実験を行い,その挙動に対して議論する. キーワード 深層学習,制約付き B...

In classical Bernoulli processes, it is assumed that a single Bernoulli experiment can be described by a precise and precisely known probability distribution. However, both of these assumptions can be relaxed. A first approach, often used in sensitivity analysis, is to drop only the second assumption: one assumes the existence of a precise distribution, but has insufficient resources to determi...

Abstract We consider an M/G/1 feedback queue with optional server vacations based on Bernoulli schedule and a single vacation policy. We assume that the server provides two types of service, type 1 with probability p1 and type 2 with probability p2 with the service times following general distribution and each arriving customer may choose either type of service. However after the completion of ...

An important technique in design of controllers for uid ow is that of compensator design. The use of compensators allows for measurement of the state at a small number of places in the uid and estimation of the state based on those measurements. However, questions arise regarding what states to measure, what locations to place sensors and how to design low order (practical) controllers. One app...